CaltechTHESIS committee: Monograph
https://feeds.library.caltech.edu/people/Burdick-J-W/combined_committee.rss
A Caltech Library Repository Feedhttp://www.rssboard.org/rss-specificationpython-feedgenenWed, 26 Jun 2024 12:51:55 -0700Applications of Surface Networks to Sampling Problems in Computer Graphics
https://resolver.caltech.edu/CaltechETD:etd-03132007-083552
Year: 1989
DOI: 10.7907/K9AS-GN82
<p>This thesis develops the theory, algorithms and data structures for adaptive sampling of parametric functions, which can represent the shapes and motions of physical objects. For the first time, ensured methods are derived for determining collisions and other interactions for a broad class of parametric functions. A new data structure, called a <i>surface network</i>, is developed for the collision algorithm and for other sampling problems in computer graphics. A surface network organizes a set of parametric samples into a hierarchy. Surface networks are shown to be good for rendering images, for approximating surfaces, and for modeling physical environments. The basic notion of a surface network is generalized to higher-dimensional problems such as collision detection. We may think of a two-dimensional network covering a three-dimensional solid, or an <i>n</i>-dimensional network embedded in a higher-dimensional space. Surface networks are applied to the problems of adaptive sampling of static parametric surfaces, to adaptive sampling of time-dependent parametric surfaces, and to a variety of applications in computer graphics, robotics, and aviation.</p>
<p>First we develop the theory for adaptive sampling of static surfaces. We explore bounding volumes that enclose static surfaces, subdivision mechanisms that adjust the sampling density, and subdivision criteria that determine where samples should be placed.</p>
<p>A new method is developed for creating bounding ellipsoids of parametric surfaces using a Lipschitz condition to place bounds on the derivatives of parametric functions. The bounding volumes are arranged in a hierarchy based on the hierarchy of the surface network. The method ensures that the bounding volume hierarchy contains the parametric surface completely. The bounding volumes are useful for computing surface intersections. They are potentially useful for ray tracing of parametric surfaces.</p>
<p>We develop and examine a variety of subdivision mechanisms to control the sampling process for parametric functions. Some of the methods are shown to improve the robustness of adaptive sampling. Algorithms for one mechanism, using bintrees of right parametric triangles, are particularly simple and robust.</p>
<p>A set of empirical subdivision criteria determine where to sample a surface, when we have no additional information about the surface. Parametric samples are concentrated in regions of high curvature, and along intersection boundaries.</p>
<p>Once the foundations of adaptive sampling for static surfaces are described, we examine time-dependent surfaces. Based on results with the empirical subdivision criteria for static surfaces, we derive ensured criteria for collision determination. We develop a new set of rectangular bounding volumes, apply a standard <i>k</i>-dimensional subdivision mechanism called k-d trees, and develop criteria for ensuring that we detect collisions between parametric surfaces.</p>
<p>We produce rectangular bounding boxes using a "Jacobian"-style matrix of Lipschitz conditions on the parametric function. The rectangular method produces even tighter bounds on the surface than the ellipsoidal method, and is effective for computing collisions between parametric surfaces.</p>
<p>A new collision determination technique is developed that can detect collisions of parametric functions, based on surface network hierarchies. The technique guarantees that the first collision is found, to within the temporal accuracy of the computation, for surfaces with bounded parametric derivatives. Alternatively, it is possible to guarantee that no collisions occur for the same class of surfaces. When a collision is found, the technique reports the location and parameters of the collision as well as the time of first collision.</p>
<p>Finally, we examine several applications of the sampling methods. Surface networks are applied to the problem of converting a two-dimensional image, or texture map, into a set of triangles that tile the plane. Many polygon-rendering systems do not provide the capability of rendering surfaces with textures. The technique converts textures to triangles that can be rendered directly by a polygon system. In addition, potential applications of the collision determination techniques are discussed, including robotics and air-traffic control problems.</p>https://resolver.caltech.edu/CaltechETD:etd-03132007-083552Applications of Surface Networks to Sampling Problems in Computer Graphics
https://resolver.caltech.edu/CaltechETD:etd-03132007-083552
Year: 1989
DOI: 10.7907/K9AS-GN82
<p>This thesis develops the theory, algorithms and data structures for adaptive sampling of parametric functions, which can represent the shapes and motions of physical objects. For the first time, ensured methods are derived for determining collisions and other interactions for a broad class of parametric functions. A new data structure, called a <i>surface network</i>, is developed for the collision algorithm and for other sampling problems in computer graphics. A surface network organizes a set of parametric samples into a hierarchy. Surface networks are shown to be good for rendering images, for approximating surfaces, and for modeling physical environments. The basic notion of a surface network is generalized to higher-dimensional problems such as collision detection. We may think of a two-dimensional network covering a three-dimensional solid, or an <i>n</i>-dimensional network embedded in a higher-dimensional space. Surface networks are applied to the problems of adaptive sampling of static parametric surfaces, to adaptive sampling of time-dependent parametric surfaces, and to a variety of applications in computer graphics, robotics, and aviation.</p>
<p>First we develop the theory for adaptive sampling of static surfaces. We explore bounding volumes that enclose static surfaces, subdivision mechanisms that adjust the sampling density, and subdivision criteria that determine where samples should be placed.</p>
<p>A new method is developed for creating bounding ellipsoids of parametric surfaces using a Lipschitz condition to place bounds on the derivatives of parametric functions. The bounding volumes are arranged in a hierarchy based on the hierarchy of the surface network. The method ensures that the bounding volume hierarchy contains the parametric surface completely. The bounding volumes are useful for computing surface intersections. They are potentially useful for ray tracing of parametric surfaces.</p>
<p>We develop and examine a variety of subdivision mechanisms to control the sampling process for parametric functions. Some of the methods are shown to improve the robustness of adaptive sampling. Algorithms for one mechanism, using bintrees of right parametric triangles, are particularly simple and robust.</p>
<p>A set of empirical subdivision criteria determine where to sample a surface, when we have no additional information about the surface. Parametric samples are concentrated in regions of high curvature, and along intersection boundaries.</p>
<p>Once the foundations of adaptive sampling for static surfaces are described, we examine time-dependent surfaces. Based on results with the empirical subdivision criteria for static surfaces, we derive ensured criteria for collision determination. We develop a new set of rectangular bounding volumes, apply a standard <i>k</i>-dimensional subdivision mechanism called k-d trees, and develop criteria for ensuring that we detect collisions between parametric surfaces.</p>
<p>We produce rectangular bounding boxes using a "Jacobian"-style matrix of Lipschitz conditions on the parametric function. The rectangular method produces even tighter bounds on the surface than the ellipsoidal method, and is effective for computing collisions between parametric surfaces.</p>
<p>A new collision determination technique is developed that can detect collisions of parametric functions, based on surface network hierarchies. The technique guarantees that the first collision is found, to within the temporal accuracy of the computation, for surfaces with bounded parametric derivatives. Alternatively, it is possible to guarantee that no collisions occur for the same class of surfaces. When a collision is found, the technique reports the location and parameters of the collision as well as the time of first collision.</p>
<p>Finally, we examine several applications of the sampling methods. Surface networks are applied to the problem of converting a two-dimensional image, or texture map, into a set of triangles that tile the plane. Many polygon-rendering systems do not provide the capability of rendering surfaces with textures. The technique converts textures to triangles that can be rendered directly by a polygon system. In addition, potential applications of the collision determination techniques are discussed, including robotics and air-traffic control problems.</p>https://resolver.caltech.edu/CaltechETD:etd-03132007-083552Model Validation for Uncertain Systems
https://resolver.caltech.edu/CaltechETD:etd-10252002-162453
Year: 1990
DOI: 10.7907/7S0Z-ZY41
NOTE: Text or symbols not renderable in plain ASCII are indicated by [...]. Abstract is included in .pdf document.
Modern robust control synthesis techniques aim at providing robustness with respect to uncertainty in the form of both additive noise and plant perturbations. On the other hand, most popular system identification methods assume that all uncertainty is in the form of additive noise. This has hampered the application of robust control methods to practical problems. This thesis begins to address this disparity by considering the connection between uncertain models and data. The model validation problem addressed here is this: given experimental, data and a model with both additive noise and normbounded perturbations, is it possible that the model could produce the observed inputoutput data? This question is reformulated as an optimization problem: what is the minimum norm noise required to account for the data and meet the constraint imposed by the perturbation uncertainty? The assumptions typically used for robust control analysis are introduced and shown to lead to a constant matrix problem. This problem is studied in detail, and bounds on the size of the required noise are developed. The dimensionality issues that arise in the consideration of the structured singular value ([...]) also arise here.
A geometric framework is used to introduce a variation on [...]. This is extended to allow the consideration of robust control analysis problems that include input and output data. The more general problem is then used to illustrate the connection between [...] and the model validation theory.
The application of the theory is illustrated by a study of a laboratory process control experiment. Typical steps in the identification of a robust control model for a physical system are discussed. It is shown, by example, how the model validation theory can be used to provide insight into the limitations of uncertain models in describing physical systems.https://resolver.caltech.edu/CaltechETD:etd-10252002-162453Model Validation for Uncertain Systems
https://resolver.caltech.edu/CaltechETD:etd-10252002-162453
Year: 1990
DOI: 10.7907/7S0Z-ZY41
NOTE: Text or symbols not renderable in plain ASCII are indicated by [...]. Abstract is included in .pdf document.
Modern robust control synthesis techniques aim at providing robustness with respect to uncertainty in the form of both additive noise and plant perturbations. On the other hand, most popular system identification methods assume that all uncertainty is in the form of additive noise. This has hampered the application of robust control methods to practical problems. This thesis begins to address this disparity by considering the connection between uncertain models and data. The model validation problem addressed here is this: given experimental, data and a model with both additive noise and normbounded perturbations, is it possible that the model could produce the observed inputoutput data? This question is reformulated as an optimization problem: what is the minimum norm noise required to account for the data and meet the constraint imposed by the perturbation uncertainty? The assumptions typically used for robust control analysis are introduced and shown to lead to a constant matrix problem. This problem is studied in detail, and bounds on the size of the required noise are developed. The dimensionality issues that arise in the consideration of the structured singular value ([...]) also arise here.
A geometric framework is used to introduce a variation on [...]. This is extended to allow the consideration of robust control analysis problems that include input and output data. The more general problem is then used to illustrate the connection between [...] and the model validation theory.
The application of the theory is illustrated by a study of a laboratory process control experiment. Typical steps in the identification of a robust control model for a physical system are discussed. It is shown, by example, how the model validation theory can be used to provide insight into the limitations of uncertain models in describing physical systems.https://resolver.caltech.edu/CaltechETD:etd-10252002-162453A method for the representation and manipulation of uncertainties in preliminary engineering design
https://resolver.caltech.edu/CaltechETD:etd-11152007-080746
Year: 1990
DOI: 10.7907/g1hs-p655
Each stage of the engineering design process, and particularly the preliminary phase, includes imprecision, stochastic uncertainty, and possibilistic uncertainty. A technique is presented by which the various levels of imprecision (where imprecision is: "uncertainty in choosing among alternatives") in the description of design elements may be represented and manipulated. The calculus of Fuzzy Sets provides the foundation of the approach. An analogous method to representing and manipulating imprecision using probability calculus is presented and compared with the fuzzy calculus technique. Extended Hybrid Numbers are then introduced to combine the effects of imprecision with stochastic and possibilistic uncertainty. Using the results, a preliminary set of metrics is proposed by which a designer can make decisions among alternative configurations in preliminary design.
In general, the hypothesis underlying the techniques described above is that making more information available than conventional approaches will enhance the decision-making capability of the designer in preliminary design. A number of elemental concepts toward this hypothesis have been formulated during the evolution of this work:
• Imprecision is a hallmark of preliminary engineering design. To carry out decisions based on the information available to the designer and on basic engineering principles, the imprecise descriptions of possible solution technologies must be formalized and quantified in some way. The application of the fuzzy calculus along with a fundamental interpretation provides a new and straight-forward means by which imprecision can be represented and manipulated.
• Besides imprecision, other uncertainties, categorized as stochastic and possibilistic, are prevalent in design, even in the early stages of the design process. Providing a method by which these uncertainties can be represented in the context of the imprecision is an important and necessary step when considering the evaluation of a design's performance. Extended Hybrid Numbers have been introduced in this work in order to couple the stochastic and possibilistic components of uncertainty with imprecision such that no information is lost in the process.
• Because of the size, coupling, and complexity of the functional requirement space in any realistic design, it is difficult to make decisions with regard to the performance of a design, even with an Extended Hybrid Number representation. Defining and utilizing metrics (or figures of merit) in the evaluation of how well a design meets the functional requirements reduces the complexity of this process. Such metrics also have merit when we begin to think of languages of design and adding the necessary pragmatics of "will a generated or proposed design satisfy the performance requirements with respect to the ever-present and unavoidable uncertainties?". These concepts form the central focus of this work. The mathematical methods presented here were developed to support and formalize these ideas.https://resolver.caltech.edu/CaltechETD:etd-11152007-080746Analog VLSI circuits for sensorimotor feedback
https://resolver.caltech.edu/CaltechETD:etd-06212007-074949
Year: 1991
DOI: 10.7907/vvye-b883
This thesis presents a design framework and circuit implementations for integrating sensory and motor processing onto very large-scale integrated (VLSI) chips. The designs consist of analog circuits that are composed of bipolar and subthreshold MOS transistors. The primary emphasis in this work is the transformation from the spatially-encoded representation found in sensory images to a scalar representation that is useful for controlling motor systems.
The thesis begins with a discussion of the aggregation of sensory signals and the resulting extraction of high-level features from sensory images. An integrated circuit that computes the centroid of a visual image is presented. A theoretical analysis of the function of this circuit in stimulus localization and a detailed error analysis are also presented. Next, the control of motors using pulse trains is discussed. Pulse-generating circuits for use in bidirectional motor control and the implementation of traditional control schemes are presented. A method for analyzing the operation of these controllers is also discussed. Finally, a framework for the combination of sensory aggregation and pulse-encoded outputs is presented. The need for signal normalization and circuits to perform this task are discussed. Two complete sensorimotor feedback systems are presented.https://resolver.caltech.edu/CaltechETD:etd-06212007-074949Generative Modeling: An Approach to High Level Shape Design for Computer Graphics and CAD
https://resolver.caltech.edu/CaltechETD:etd-07122007-144802
Year: 1991
DOI: 10.7907/HRFJ-QC74
<p>Generative modeling is an approach to computer-assisted geometric modeling. The goal of the approach is to allow convenient and high-level specification of shapes, and provide tools for rendering and analysis of the specified shapes. Shapes include curves, surfaces, and solids in 3D space, as well as higher-dimensional entities such as surfaces deforming in time, and solids with a spatially varying mass density.</p>
<p>Shape specification in the approach involves combining low-dimensional entities, especially 2D curves, into higher-dimensional shapes. This combination is specified through a powerful shape description language which builds multidimensional parametric functions. The language is based on a set of primitive operators on parametric functions which include arithmetic operators, vector and matrix operators, integration and differentiation, constraint solution and global optimization. Although each-primitive operator is fairly simple, high-level shapes and shape building operators can be defined using recursive combination of the primitive operators.</p>
<p>The approach encourages the modeler to build parameterized families of shapes rather than single instances. Shapes can be parameterized by scalar parameters (e.g., time or joint angle) or higher-dimensional parameters (e.g., a curve controlling how the scale of a cross section varies as it is translated). Such parameterized shapes allow easy modification of the design, since the modeler can interact with parameters that relate to high-level properties of the shape. In contrast, many geometric modeling systems use a much lower-level specification, such as through sets of many 3D control points.</p>
<p>Tools for rendering and analysis of generative models are developed using the concept of interval analysis. Each primitive operator on parametric functions has an inclusion function method, which produces an interval bound on the range of the function, given an interval bound on its domain. With these inclusion functions, robust algorithms exist for computing solutions to nonlinear systems of constraints and global minimization problems, when these problems are expressed in the modeling language. These algorithms, in turn, are developed into robust approximation techniques to compute intersections, CSG operations, and offset operations.</p>https://resolver.caltech.edu/CaltechETD:etd-07122007-144802Robustness properties of nonlinear process control and implications for the design and control of a packed bed reactor
https://resolver.caltech.edu/CaltechETD:etd-07112007-084012
Year: 1991
DOI: 10.7907/3n52-hk32
The robustness properties of nonlinear process control are studied with particular emphasis on applications to the design and control of a catalytic fixed bed reactor.
Analysis tools are developed to determine the stability and performance of nonlinear dynamical systems. The results are based upon new extensions of the structured singular value to a class of nonlinear and time-varying systems. Conic sectors are utilized in approximating the static nonlinearities present and an algorithm is developed for optimal conic sector calculation.
The synthesis tools of differential geometry are studied with respect to their closed loop robust performance properties. New results in approximate linearization are contrasted with exact linearization and linear control. It is shown that the approximate linearization technique is superior with respect to disturbance handling, optimization of the resultant transformations, and range of applicability.
Nonlinear approaches for the control of a packed bed reactor are investigated. In particular, the differential geometric technique of input-output linearization is found to yield superior closed-loop performance over regions of open-loop parametric sensitivity. The synthesis of a linearizing controller for this nonlinear distributed parameter system involves a two-tier approach. In the first stage, a low order nonlinear model is developed for the reactor. This is accomplished by treating the active transport mechanisms in the bed as a nonlinear wave which propagates through the bed in response to changes in the operating conditions. The resultant lumped parameter model facilitates the design of the input-output linearizing controller in the second tier of this scheme. The implementational hurdles for this approach are identified and comparisons are drawn on the strengths of this approach over robust linear control for the reactor.
Practical guidelines are developed for the design of packed bed reactors. The criteria result from requirements on the radial temperature profile, temperature sensitivity, and acceptable pressure drop. The stabilizing effects of feedback control for industrial fixed bed catalytic reactors are addressed. Simulations support the result that violation of the proposed criteria leads to unacceptable closed-loop performance.
In conclusion, general guidelines are constructed from a series of case studies on the proper selection of linear versus "linearizing" control. The relative performance is measured by the region of attraction, magnitude of manipulated variable action, and sensitivity to input disturbances. The work represents the first objective evaluation of the strengths and limitations of input-output linearization compared to linear control.https://resolver.caltech.edu/CaltechETD:etd-07112007-084012Robust adaptive control of manipulators with application to joint flexibility
https://resolver.caltech.edu/CaltechETD:etd-08072007-073507
Year: 1992
DOI: 10.7907/ks5z-8e20
This thesis discusses the model-based adaptive trajectory control of commercial manipulators whose dynamics are well known with uncertainties confined to parameters.
This thesis emphasizes the importance of the transient behavior as well as robust stability of a system and takes it into account in the design of adaptive control laws. The basic idea is to search for compensators in the direction of minimizing a quadratic performance index, and then analyze the stability and robustness of the selected compensators in the presence of bounded disturbances, sensor noises, and unmodelled dynamics. With this idea, centralized and decentralized adaptive control schemes are proposed for rigid-joint manipulators. Stability bounds for disturbances, control and adaptation gains, and desired trajectories and their time-derivatives are derived for the proposed schemes. These bounds are sufficient conditions for robust stability of the proposed schemes in the presence of unmodelled dynamics such as feedback delays in the digital control systems and the coupled dynamics in the decentralized scheme.
A flexibility compensator is designed to treat the problem of joint flexibility. With the flexibility compensator, a manipulator having flexible joints is transformed to that having rigid joints with high-frequency dynamics of joint couplings representing unmodelled dynamics. In this way, control of flexible-joint manipulators is converted to that of the corresponding rigid-joint manipulators. Accordingly, the robust adaptive control schemes proposed for rigid-joint manipulators are applied. Then, through stability analysis, stability bounds for disturbances, control and adaptation gains, and desired trajectories and their time-derivatives are derived for the scheme with the flexibility compensator, in the presence of the unmodelled dynamics. Under the constraint of these bounds, the proposed adaptive scheme is not only almost independent of the gear-reduction ratios, flexibilities of joint couplings, and characteristics of actuators, but also free from the requirements of measuring angular accelerations and jerks of links.https://resolver.caltech.edu/CaltechETD:etd-08072007-073507Theory and Applications of Hyper-Redundant Robotic Manipulators
https://resolver.caltech.edu/CaltechETD:etd-11082006-132210
Year: 1992
DOI: 10.7907/F12D-0X25
The term "hyper-redundant" refers to robotic manipulators and mobile robots with a very large, possibly infinite, number of actuatable degrees of freedom. These robots are analogous in morphology and operation to snakes, worms, elephant trunks, and tentacles. This thesis presents a novel kinematic framework for hyper-redundant manipulator motion planning and task implementation. The basis of this formulation is the use of a "backbone reference set" which captures the essential macroscopic geometric features of hyper-redundant robots. In the analytical part of this work, the backbone representation is developed and used to solve problems in obstacle avoidance, locomotion, grasping, and "optimal" end effector placement. The latter part of this thesis deals with the design and implementation of a thirty-degree-of-freedom planar hyper-redundant manipulator which is used to demonstrate these novel kinematic and motion planning techniques. Design issues such as robustness with respect to mechanical failure, and design for easy assembly and repair are also addressed. The analytical and design concepts are combined to illustrate tasks for which hyper-redundant robotic mechanisms are well suited.
https://resolver.caltech.edu/CaltechETD:etd-11082006-132210A structured approach to physically-based modeling for computer graphics
https://resolver.caltech.edu/CaltechTHESIS:09282011-075406850
Year: 1992
DOI: 10.7907/tbgd-g285
<p>This thesis presents a framework for the design of physically-based computer graphics models. The framework includes a paradigm for the structure of physically-based models, techniques for "structured" mathematical modeling, and a specification of a computer program structure in which to implement the models. The framework is based on known principles and methodologies of structured programming and mathematical modeling. Because the framework emphasizes the structure and organization of models, we refer to it as "Structured Modeling."</p>
<p>The Structured Modeling framework focuses on clarity and "correctness" of models, emphasizing explicit statement of assumptions, goals, and techniques. In particular, we partition physically-based models, separating them into conceptual and mathematical models, and posed problems. We control complexity of models by designing in a modular manner, piecing models together from smaller components.</p>
<p>The framework places a particular emphasis on defining a complete formal statement of a model's mathematical equations, before attempting to simulate the model. To manage the complexity of these equations, we define a collection of mathematical constructs, notation, and terminology, that allow mathematical models to be created in a structured and modular manner.</p>
<p>We construct a computer programming environment that directly supports the implementation of models designed using the above techniques. The environment is geared to a tool-oriented approach, in which models are built from an extensible collection of software objects, that correspond to elements and tasks of a "blackboard" design of models.</p>
<p>A substantial portion of this thesis is devoted to developing a library of physically-based model "modules," including rigid-body kinematics, rigid-body dynamics, and dynamic constraints, all built with the Structured Modeling framework. These modules are intended to serve both as examples of the framework, and as potentially useful tools for the computer graphics community. Each module includes statements of goals and assumptions, explicit mathematical models and problem statements, and descriptions of software objects that support them. We illustrate the use of the library to build some sample models, and include discussion of various possible additions and extensions to the library.</p>
<p>Structured Modeling is an experiment in modeling: an exploration of designing via strict adherence to a dogma of structure, modularity, and mathematical formality. It does not stress issues such as particular numerical simulation techniques or efficiency of computer execution time or memory usage, all of which are important practical considerations in modeling. However, at least so far as the work carried on in this thesis, Structured Modeling has proven to be a useful aid in the design and understanding of complex physically based models.</p>https://resolver.caltech.edu/CaltechTHESIS:09282011-075406850Robotic hand-eye motor learning
https://resolver.caltech.edu/CaltechETD:etd-12052007-130729
Year: 1993
DOI: 10.7907/47t1-2f31
This thesis investigates the use of neural networks and nonlinear estimation in robotic motor learning. It presents a detailed experimental investigation of the performance and parametric sensitivity of resource-allocating neural networks along with a new learning algorithm that offers rapid adaptation and excellent accuracy. It also includes an appendix that relates feed-forward neural networks to familiar mathematical ideas.
In addition, it presents two learning hand-eye calibration systems, one based on neural networks and the other on nonlinear estimation. The network-based system learns to correct robot positioning errors arising from the use of nominal system kinematics, while the estimation-based system identifies the robot's kinematic parameters. Both systems employ the same two-link robot with stereo vision, and include noise and various other error sources. The network-based system is robust to all error sources considered, though noise naturally limits performance. The estimation-based system has significantly better performance when the robot and vision systems are well modeled, but is extremely sensitive to unmodeled error sources and noise.
Finally, it presents a robot control system based on neural networks that learns to catch balls perfectly without requiring explicit programming or conventional controllers. It uses only feed-forward pursuit motions learned through practice, and is initially incapable of even moving its arm in response to external stimuli. It learns to identify and control its pursuit movements, to identify and predict ball behavior, and, with the aid of advice from a critic, to modify its movement commands to improve catching success. The system, which incorporates information from visual, arm state, and drive force sensors, characterizes control situations using input/response pairs. This allows it to learn and respond to plant variations without requiring parametric models or parameter identification. It achieves robust execution by comparing predicted and observed behavior, using inconsistencies to trigger learning and behavioral change. The architectural approach, which involves both declarative and analog knowledge as well as short- and long-term memory, can be extended to learning other sensor-motor skills like mechanical assembly and synchronizing motor actions with external processes.https://resolver.caltech.edu/CaltechETD:etd-12052007-130729Robotic hand-eye motor learning
https://resolver.caltech.edu/CaltechETD:etd-12052007-130729
Year: 1993
DOI: 10.7907/47t1-2f31
This thesis investigates the use of neural networks and nonlinear estimation in robotic motor learning. It presents a detailed experimental investigation of the performance and parametric sensitivity of resource-allocating neural networks along with a new learning algorithm that offers rapid adaptation and excellent accuracy. It also includes an appendix that relates feed-forward neural networks to familiar mathematical ideas.
In addition, it presents two learning hand-eye calibration systems, one based on neural networks and the other on nonlinear estimation. The network-based system learns to correct robot positioning errors arising from the use of nominal system kinematics, while the estimation-based system identifies the robot's kinematic parameters. Both systems employ the same two-link robot with stereo vision, and include noise and various other error sources. The network-based system is robust to all error sources considered, though noise naturally limits performance. The estimation-based system has significantly better performance when the robot and vision systems are well modeled, but is extremely sensitive to unmodeled error sources and noise.
Finally, it presents a robot control system based on neural networks that learns to catch balls perfectly without requiring explicit programming or conventional controllers. It uses only feed-forward pursuit motions learned through practice, and is initially incapable of even moving its arm in response to external stimuli. It learns to identify and control its pursuit movements, to identify and predict ball behavior, and, with the aid of advice from a critic, to modify its movement commands to improve catching success. The system, which incorporates information from visual, arm state, and drive force sensors, characterizes control situations using input/response pairs. This allows it to learn and respond to plant variations without requiring parametric models or parameter identification. It achieves robust execution by comparing predicted and observed behavior, using inconsistencies to trigger learning and behavioral change. The architectural approach, which involves both declarative and analog knowledge as well as short- and long-term memory, can be extended to learning other sensor-motor skills like mechanical assembly and synchronizing motor actions with external processes.https://resolver.caltech.edu/CaltechETD:etd-12052007-130729High-resolution optoelectronic and photogrammetic 3-D surface geometry acquisition and analysis
https://resolver.caltech.edu/CaltechETD:etd-08292007-091850
Year: 1993
DOI: 10.7907/wn0s-6y22
A high-resolution, high-speed, automatic, and non-contact 3-D surface geometry measuring system has been developed. It is based on a photogrammetric and optoelectronic technique that adopts lateral-photoeffect diode detectors sensitive in the near-infrared range. Two cameras in stereo positions are both equipped with the large 2-axis analog detectors. A light beam is focused and scanned onto the surface of an object as a very small light spot. Excitations on detectors generated by the reflected light from the spot create photocurrents that are transformed into 2-D position signals in a very short time. A simple set of calculations is done to photogrammetrically triangulate two sets of 2-D coordinates from the detectors into the 3-D coordinates of the light spot. Because only one small light spot in the scene is illuminated at a time, the stereo-correspondence problem is solved in real time. The detectors are able to collect data at 10 KHz with 4,096x4,096 resolution based on a 12-bit A/D converter. The resolution and precision can be improved up to eight times by oversampling. The system is able to resolve, for example, less than 10 µm from 47 cm away with a nominal viewing volume of (22 cm)[superscript 3]. Its performance is better than contemporary coordinate measuring, range finding, shape digitizing, and machine vision systems, and is comparable to the best aspects of each existing system. The irregular 3-D data it generates can be regularized so that data processing algorithms designed for image systems may be applied. The system is designed for the acquisitions of general surface geometries, such as fabricated parts, machined surfaces, biological surfaces, and deformed parts. The system will be useful in solving a variety of 3-D surface geometry measuring problems in engineering design, manufacturing, inspection, robot kinematics measurement, and vision.
https://resolver.caltech.edu/CaltechETD:etd-08292007-091850Cavitation and wake structure of unsteady tip vortex flows
https://resolver.caltech.edu/CaltechETD:etd-03272007-131947
Year: 1993
DOI: 10.7907/ANNN-VC25
NOTE: Text or symbols not renderable in plain ASCII are indicated by [...]. Abstract is included in .pdf document.
Unsteady flows are prevalent in virtually every fluid application yet, because of their intrinsic complexity, few attempts have been made to measure them or explain their behavior. This thesis presents an experimental study of one of the simplest unsteady flow induced effects, the periodic change in angle of attack of a lifting surface. Of particular interest is the influence this effect has on the tip vortex structure of a finite aspect ratio hydrofoil and the part it plays in the inception of cavitation.
An aspect ratio 2.3 hydrofoil was reflection-plane mounted to the test section floor of the Caltech Low Turbulence Water Tunnel and harmonically oscillated in pitch near its center of pressure. Observations of the growth and collapse of surface and tip vortex cavitation were made along with detailed observations of the interaction of the tip vortex formation with the spanwise wake structure. Measurements of the cavitation inception number for surface cavitation and tip vortex cavitation were made relative to the phase of the hydrofoil and the reduced frequency, k=[low-case omega]c/2U[...], of oscillation. Studies of the oscillation-induced spanwise trailing vortex structures and the Karman vortex street generated by the boundary layer were made of a two-dimensional hydrofoil. Laser Doppler Velocimetry (LDV) measurements were taken of the tip vortex velocity profile and the flow at the trailing edge of both the two-and the three-dimensional hydrofoils at reduced frequencies ranging from 0.5 to 2.0. Dynamic changes in bound circulaion and shed vorticity in the streamwise and spanwise directions relative to the freestream were calculated from these measurements at three locations along the span of the foil. The results of these measurements are compared to theoretical flow calculations and related to measurements of the cavitation inception number in the tip vortex region of the three-dimensional foilhttps://resolver.caltech.edu/CaltechETD:etd-03272007-131947Theory and Applications of Modular Reconfigurable Robotic Systems
https://resolver.caltech.edu/CaltechETD:etd-10202005-090745
Year: 1994
DOI: 10.7907/2AAA-RY45
A modular reconfigurable robotic system consists of various link and joint units with standardized connecting interfaces that can be easily separated and reassembled into different configurations. Compared to a fixed configuration robot, which is usually a compromised design for a limited set of tasks, a modular robot can accomplish a large class of tasks through reconfiguration of a small inventory of modules. This thesis studies how to find an optimal module assembly configuration constructed from a given inventory of module components for a specific task. A set of generalized module models that bear features found in many real implementations is introduced. The modular robot assembly configuration is represented by a novel Assembly Incidence Matrix (AIM). Equivalence relations based on module geometry symmetries and graph isomorphisms are defined on the AIMs. An enumeration algorithm to generate non-isomorphic assembly configurations based on this equivalence relation is proposed. Examples demonstrate that this method is a significant improvement over a brute force enumeration process. Configuration independent kinematic models for modular robots are developed, and they are essential for solving the task-optimal configuration problem. A task-oriented objective function is defined on the set of non-isomorphic module assembly configurations. Task requirements and kinematic constraints on the robot assembly are treated as parameters to this objective function. The task-optimal configuration problem is formulated as a combinatorial optimization problem to which genetic algorithms are employed for solutions. Examples of finding task-optimal serial revolute-jointed robot configurations are demonstrated. In addition, the applications of modular robots to planning multifinger grasping and manipulation are developed. Planning two-finger grasps is done through finding antipodal point grasps on smooth shaped objects. Planning n-finger grasps is achieved by defining a qualitative force-closure test function on the n-finger grasps on an object. Applications of this test function to manipulation task and finger gaiting are illustrated.https://resolver.caltech.edu/CaltechETD:etd-10202005-090745Aspects of geometric mechanics and control of mechanical systems
https://resolver.caltech.edu/CaltechETD:etd-07132006-143133
Year: 1995
DOI: 10.7907/CHWF-M421
Many interesting control systems are mechanical control systems. In spite of this, there has not been much effort to develop methods which use the special structure of mechanical systems to obtain analysis tools which are suitable for these systems. In this dissertation we take the first steps towards a methodical treatment of mechanical control systems.
First we develop a framework for analysis of certain classes of mechanical control systems. In the Lagrangian formulation we study "simple mechanical control systems" whose Lagrangian is "kinetic energy minus potential energy." We propose a new and useful definition of controllability for these systems and obtain a computable set of conditions for this new version of controllability. We also obtain decompositions of simple mechanical systems in the case when they are not controllable. In the Hamiltonian formulation we study systems whose control vector fields are Hamiltonian. We obtain decompositions which describe the controllable and uncontrollable dynamics. In each case, the dynamics are shown to be Hamiltonian in a suitably general sense.
Next we develop intrinsic descriptions of Lagrangian and Hamiltonian mechanics in the presence of external inputs. This development is a first step towards a control theory for general Lagrangian and Hamiltonian control systems. Systems with constraints are also studied. We first give a thorough overview of variational methods including a comparison of the "nonholonomic" and "vakonomic" methods. We also give a generalised definition for a constraint and, with this more general definition, we are able to give some preliminary controllability results for constrained systems.
https://resolver.caltech.edu/CaltechETD:etd-07132006-143133Exponential stabilization of driftless nonlinear control systems
https://resolver.caltech.edu/CaltechETD:etd-10172007-104556
Year: 1995
DOI: 10.7907/7myb-h217
NOTE: Text or symbols not renderable in plain ASCII are indicated by [...]. Abstract is included in .pdf document.
This dissertation lays the foundation for practical exponential stabilization of driftless control systems. Driftless systems have the form,
[...].
Such systems arise when modeling mechanical systems with nonholonomic constraints. In engineering applications it is often required to maintain the mechanical system around a desired configuration. This task is treated as a stabilization problem where the desired configuration is made an asymptotically stable equilibrium point. The control design is carried out on an approximate system. The approximation process yields a nilpotent set of input vector fields which, in a special coordinate system, are homogeneous with respect to a non-standard dilation. Even though the approximation can be given a coordinate-free interpretation, the homogeneous structure is useful to exploit. Since implementing a controller requires choosing a coordinate system, there are extra benefits to be gained by choosing coordinates in which the approximation is homogeneous. The feedbacks are required to be homogeneous functions and thus preserve the homogeneous structure in the closed-loop system. The stability achieved is called p-exponential stability. This extended notion of exponential stability is required since the feedback, and hence the closed-loop system, is not Lipschitz. However, it is shown that the convergence rate of a Lipschitz closed-loop driftless system cannot be bounded by an exponential envelope.
The synthesis methods generate feedbacks which are not smooth on [...]. The solutions of the closed-loop system are proven to be unique in this case. In addition, for many driftless systems the control inputs are often velocities. A more appropriate formulation of the stabilization problem has the control law specifying forces instead of velocities. We have extended the kinematic velocity controllers to controllers which command forces and still p-exponentially stabilize the system.
Perhaps the ultimate justification of the methods proposed in this thesis are the experimental results. The experiments demonstrate the superior convergence performance of the p-exponential stabilizers versus traditional smooth feedbacks. The experiments also highlight the importance of transformation conditioning in the feedbacks. Other design issues, such as scaling the measured states to eliminate hunting, are discussed. The methods and problems in this thesis bring the practical control of strongly nonlinear systems one step closer.
https://resolver.caltech.edu/CaltechETD:etd-10172007-104556Mode-Like Properties and Identification of Nonlinear Vibrating Systems
https://resolver.caltech.edu/CaltechTHESIS:12012011-113838877
Year: 1995
DOI: 10.7907/707b-jf67
<p>A study is made of mode-like properties and identification of nonlinear systems and their applications in structural seismic analysis.</p>
<p>In the thesis, mode-like behavior of nonlinear systems is examined. The modal frequencies and mode shapes of nonlinear systems are found to be dependent on the
response. Based on approximation, amplitude-dependent mode shape is defined and approximate methods for calculation of modal frequencies and mode shapes (instantaneous and amplitude-dependent) are presented. Based on amplitude-dependent modal relationship, amplitude-dependent models of modal equations which are valid in large range of response
and suitable for unique identification are proposed and the corresponding modal identification procedures are developed. The applicability of the new models and
identification algorithms is tested through the analysis of an ideal 3DOF nonlinear system.</p>
<p>As applications, the seismic responses of a 47-story building and a 4-story building are investigated using the presented methods. The modal parameters and modal equations
of the structures are identified.</p>https://resolver.caltech.edu/CaltechTHESIS:12012011-113838877A geometric framework for dynamic vision
https://resolver.caltech.edu/CaltechETD:etd-01082008-103705
Year: 1996
DOI: 10.7907/x87w-t943
This thesis explores the problem of inferring information about the three-dimensional world from its projections onto a camera (images). Among all visual cues, we do not address "pictorial" ones, such as texture or shading. Instead, we concentrate on "dynamic" cues, which are associated with variations of the image over time.
In order to eliminate pictorial cues, one may represent the world as a collection of geometric primitives, such as points, curves or surfaces in three-dimensional space. Then, from the two-dimensional motion of the projection of such primitives onto the image, one can infer the three-dimensional structure of the world and its motion relative to the viewer.
"Three-dimensional structure from two-dimensional images" has now been a central theme in Computer Vision for over two decades, and tools from Linear Algebra and Projective Geometry have been widely employed to attack the problem as a "static" task. It is only in recent years that the role of time has started to be recognized, after the influential work of Dickmanns and his coworkers on vehicle guidance on freeways.
We do not impose restrictions on the structure of the environment, and we cast the problem of general three-dimensional structure and motion estimation within the framework of Dynamical Systems. We show how different algebraic constraints on the image projections can be interpreted as nonlinear and implicit dynamical models whose (unknown) parameters live in peculiar differentiable manifolds that encode three-dimensional information. Recovering such three-dimensional information then amounts to identifying dynamical models while taking into account the geometry of the parameter manifolds.https://resolver.caltech.edu/CaltechETD:etd-01082008-103705Model Validation, Control, and Computation
https://resolver.caltech.edu/CaltechETD:etd-01032008-090000
Year: 1996
DOI: 10.7907/nvwd-hw47
<p>Engineering in general is concerned with controlling and predicting future behavior with some certainty despite having only imperfect information. Although feedback can be an exceptionally effective engineering tool and is often easy to apply, the behavior of a system under feedback can be extremely sensitive to model mismatch, which is always present. The potential for unpredictable behavior is a major drawback to the engineering application of feedback. Robust control theory addresses this difficulty by parametrizing a family of feedback controllers that are less sensitive to model mismatch.</p>
<p>Despite encouraging early applications, robust control theory has so far been deficient in analysis of systems, synthesis of controllers, connection to real problems, and applicability to nonlinear problems. Further, results on the computational complexity of robust control problems that necessitate either bounds computation or a restricted class of problems have cast doubts about the potential utility of the area.</p>
<p>Initial work in robust control focused on complex uncertainty in the frequency domain. A perceived deficiency is that such model sets are unrealistic: uncertainty in mass, stiffness, aero-coefficients, and the like are naturally modeled as real variations. This thesis includes initial work on practical upper bound computation and substantially improved lower bound computation for moderately large robust control analysis problems that include such real parametric uncertainty, despite the computational complexity of the problems. Although better upper bound computation than that described here is now available for small problems, such is not the case for large problems. The improved lower bound computation chronicled here is desirable because the initial lower bound computation for problems with real parametric uncertainty is not as reliable as in the complex case. Additionally, this thesis shows that branch and bound is a limited but critical tool for better computation, a fact that previously has gone unrecognized.</p>
<p>Together, these contributions allow for the practical computation of robust control problems of engineering interest and provide the basis not only for applications that may ultimately determine the utility of the robust control paradigm but also for the computation of various outgrowths of the [mu] framework, which is the basis for computational robust control.</p>
<p>One such outgrowth is the model validation problem. Model validation tests whether a robust control model in the [mu] framework is consistent with experimentally determined time histories quite a different problem than standard system identification. This thesis shows that the model validation problem is indeed closely related to the standard [mu] problem and its computation.</p>
<p>The practical computation of the model validation problem, which should follow naturally from the work presented here, provides the basis for the connection between robust control theory and practical applications. Future work along these lines should elevate the application of robust control theory from chance and intuition to a standard engineering tool.</p>
<p>Further, the techniques that render the model validation problem similar to the standard [mu] problem are applicable to a great variety of systems analysis and design problems. This newly perceived generality of the [mu] paradigm may ultimately provide a unifying framework for the many seemingly disparate aspects of systems and control design.</p>https://resolver.caltech.edu/CaltechETD:etd-01032008-090000Experimental Control: a Helicopter Case Study
https://resolver.caltech.edu/CaltechETD:etd-12202007-115753
Year: 1996
DOI: 10.7907/sdad-de68
<p>Robust control has not been used as widely as it could because modelling tools have not advanced as far as analysis and synthesis tools. This becomes readily apparent when applying robust control theory to real problems. With this in mind, an experimental platform was designed and built to study the application of robust control. This platform consists of a real-time computer and a radio-controlled model helicopter mounted on a six degree-of-freedom stand. Experimental systems provide the opportunity not only to verify the applicability of new control theory but also to highlight potential deficiencies.</p>
<p>Traditional system identification and control techniques were used to construct hover controllers for the model helicopter. These techniques are not suitable for the construction of robust models for a system of this complexity. In particular, there was no systematic way to augment nominal identified models with uncertainty suitable for the construction of robust controllers.</p>
<p>To address this issue, frequency-domain model validation algorithms and software were developed. These algorithms provide a methodology for verifying the applicability and consistency between experimental data and robust models. Additionally, they provide a method whereby nominal model parameters can be tuned in a robust setting. This is the first set of software tools which provide this capability for general linear uncertain systems.</p>
<p>Using these new software tools, a systematic design process was developed which incorporated frequency-domain model validation analysis, µ-analysis and µ-synthesis, simulation, and implementation. This design process proved to be a valuable new tool for constructing robust models and designing robust control systems. In particular, by applying this design process to the helicopter, the size of uncertainty in the robust model was substantially reduced without sacrificing the ability of the model to "cover" experimental data and the first controller implemented performed well. This was strikingly different from the results obtained when using standard robust control techniques, where several controllers destabilized the helicopter when implemented, even though they performed well under simulation.</p>
<p>The model validation software and design process provide a consistent methodology and systematic framework which connects system identification, the construction of robust models, and controller synthesis with experimental data. For the first time the control engineer can compute measures on the validity of a robust model, with respect to all observed data on the actual physical system, which are directly related to the robustness measures resulting from µ-analysis and µ-synthesis.</p>
https://resolver.caltech.edu/CaltechETD:etd-12202007-115753Sensor Based Motion Planning: The Hierarchical Generalized Voronoi Graph
https://resolver.caltech.edu/CaltechETD:etd-12182007-090504
Year: 1996
DOI: 10.7907/49ee-a204
Sensor based motion planning incorporates sensor information reflecting the state of a robot's environment into its planning process, whereas traditional approaches assume complete prior knowledge of the robot's environment. Recent research has focused on the development and incremental construction of the hierarchical generalized Voronoi graph (HGVG), which is a concise representation of a robot's environment. The HGVG is advantageous in that it lends itself to sensor based construction in a rigorous and provably correct manner. With this approach, a robot can enter an unknown environment, incrementally construct the HGVG, and then use the HGVG for future excursions in the environment. Simulations and experiments validate this approach.https://resolver.caltech.edu/CaltechETD:etd-12182007-090504The mechanics and control of undulatory robotic locomotion
https://resolver.caltech.edu/CaltechETD:etd-10202005-153514
Year: 1996
DOI: 10.7907/Y1TF-RF86
In this dissertation, we examine a formulation of problems of undulatory robotic locomotion within the context of mechanical systems with nonholonomic constraints and symmetries. Using tools from geometric mechanics, we study the underlying structure found in general problems of locomotion. In doing so, we decompose locomotion into two basic components: internal shape changes and net changes in position and orientation. This decomposition has a natural mathematical interpretation in which the relationship between shape changes and locomotion can be described using a connection on a trivial principal fiber bundle.
We begin by reviewing the processes of Lagrangian reduction and reconstruction for unconstrained mechanical systems with Lie group symmetries, and present new formulations of this process which are easily adapted to accommodate external constraints. Additionally, important physical quantities such as the mechanical connection and reduced mass-inertia matrix can be trivially determined using this formulation. The presence of symmetries then allows us to reduce the necessary calculations to simple matrix manipulations.
The addition of constraints significantly complicates the reduction process; however, we show that for invariant constraints, a meaningful connection can be synthesized by defining a generalized momentum representing the momentum of the system in directions allowed by the constraints. We then prove that the generalized momentum and its governing equation possess certain invariances which allows for a reduction process similar to that found in the unconstrained case. The form of the reduced equations highlights the synthesized connection and the matrix quantities used to calculate these equations.
The use of connections naturally leads to methods for testing controllability and aids in developing intuition regarding the generation of various locomotive gaits. We present accessibility and controllability tests based on taking derivatives of the connection, and relate these tests to taking Lie brackets of the input vector fields.
The theory is illustrated using several examples, in particular the examples of the snakeboard and Hirose snake robot. We interpret each of these examples in light of the theory developed in this thesis, and examine the generation of locomotive gaits using sinusoidal inputs and their relationship to the controllability tests based on Lie brackets.
https://resolver.caltech.edu/CaltechETD:etd-10202005-153514A geometric framework for dynamic vision
https://resolver.caltech.edu/CaltechETD:etd-01082008-103705
Year: 1996
DOI: 10.7907/x87w-t943
This thesis explores the problem of inferring information about the three-dimensional world from its projections onto a camera (images). Among all visual cues, we do not address "pictorial" ones, such as texture or shading. Instead, we concentrate on "dynamic" cues, which are associated with variations of the image over time.
In order to eliminate pictorial cues, one may represent the world as a collection of geometric primitives, such as points, curves or surfaces in three-dimensional space. Then, from the two-dimensional motion of the projection of such primitives onto the image, one can infer the three-dimensional structure of the world and its motion relative to the viewer.
"Three-dimensional structure from two-dimensional images" has now been a central theme in Computer Vision for over two decades, and tools from Linear Algebra and Projective Geometry have been widely employed to attack the problem as a "static" task. It is only in recent years that the role of time has started to be recognized, after the influential work of Dickmanns and his coworkers on vehicle guidance on freeways.
We do not impose restrictions on the structure of the environment, and we cast the problem of general three-dimensional structure and motion estimation within the framework of Dynamical Systems. We show how different algebraic constraints on the image projections can be interpreted as nonlinear and implicit dynamical models whose (unknown) parameters live in peculiar differentiable manifolds that encode three-dimensional information. Recovering such three-dimensional information then amounts to identifying dynamical models while taking into account the geometry of the parameter manifolds.https://resolver.caltech.edu/CaltechETD:etd-01082008-103705Autonomous Reorientation of a Maneuver-Limited Spacecraft Under Simple Pointing Constraints
https://resolver.caltech.edu/CaltechETD:etd-01162008-152619
Year: 1997
DOI: 10.7907/kap2-6n63
This report presents techniques for using discrete finite rotations to reorient a spacecraft from a given initial attitude to a final attitude which satisfies a specified aiming objective. The objective may be a fully specified final orientation or it may require that the spacecraft direct an instrument along a certain direction. Constraints are also imposed on the allowable intermediate orientations that the spacecraft may assume during the course of the maneuver, representing the operational requirements of onboard instrumentation. The algorithms presented consider solutions that will achieve the desired objective with only one or two slew maneuvers, although they may be easily extended to consider more complicated solutions requiring additional maneuvers.https://resolver.caltech.edu/CaltechETD:etd-01162008-152619Autonomous Reorientation of a Maneuver-Limited Spacecraft Under Simple Pointing Constraints
https://resolver.caltech.edu/CaltechETD:etd-01162008-152619
Year: 1997
DOI: 10.7907/kap2-6n63
This report presents techniques for using discrete finite rotations to reorient a spacecraft from a given initial attitude to a final attitude which satisfies a specified aiming objective. The objective may be a fully specified final orientation or it may require that the spacecraft direct an instrument along a certain direction. Constraints are also imposed on the allowable intermediate orientations that the spacecraft may assume during the course of the maneuver, representing the operational requirements of onboard instrumentation. The algorithms presented consider solutions that will achieve the desired objective with only one or two slew maneuvers, although they may be easily extended to consider more complicated solutions requiring additional maneuvers.https://resolver.caltech.edu/CaltechETD:etd-01162008-152619The mechanics and control of robotic locomotion with applications to aquatic vehicles
https://resolver.caltech.edu/CaltechETD:etd-08122005-152639
Year: 1998
DOI: 10.7907/50M3-1529
This work illuminates the utility of a theory of locomotion rooted in geometric mechanics and nonlinear control. We regard the internal configuration of a deformable body, together with its position and orientation in ambient space, as a point in a trivial principal fiber bundle over the manifold of body deformations. We obtain connections on such bundles which describe the nonholonomic constraints, conservation laws, and force balances to which certain propulsors are subject, and contruct and analyze control-affine normal forms for different classes of systems. We examine the applicability of results involving geometric phases to the practical computation of trajectories for systems described by single connections. We propose a model for planar carangiform swimming based on reduced Euler-Lagrange equations for the interaction of a rigid body and an incompressible fluid, accounting for the generation of thrust due to vortex shedding through controlled coupling terms. We investigate the correct form of this coupling experimentally with a robotic propulsor, comparing its observed behavior with that predicted numerically.
https://resolver.caltech.edu/CaltechETD:etd-08122005-152639Control of stratified systems with robotic applications
https://resolver.caltech.edu/CaltechETD:etd-01232008-144001
Year: 1998
DOI: 10.7907/49h9-q898
Many interesting and important control systems evolve on stratified configuration spaces. Roughly speaking, a configuration manifold is called "stratified" if it contains subspaces (submanifolds) upon which the system had different equations of motion. Robotic systems, in particular, are of this nature. For example, a legged robot has discontinuous equations of motion near points in the configuration space where each of its "feet" comes into contact with the ground. In such a case, when the system moves from one submanifold to another, the equations of motion change in a non-smooth, or even discontinuous manner. In such cases, traditional nonlinear control methodologies are inapplicable because they generally rely upon some form of differentiation. Yet, it is precisely the discontinuous nature of such systems that is often their most important characteristic.
This dissertation presents methods which consider the intrinsic physical geometric structure present in such problems to address nonlinear controllability and motion planning for stratified systems. For both problems, by exploiting this geometric structure of stratified systems, we can extend standard nonlinear control results and methodologies to the stratified case. A related problem addressed by this dissertation is that of controllability of systems where some control inputs are constrained to be non-negative. This problem arises in stratified systems which arise by way of physical contact because the normal force between contacting systems must be nonnegative. For all the results, a basic goal is to generate results which are general. For example, for robotics applications, these results are independent of a particular robot's number of legs, fingers or morphology.
https://resolver.caltech.edu/CaltechETD:etd-01232008-144001Mechanics and Planning of Workpiece Fixturing and Robotic Grasping
https://resolver.caltech.edu/CaltechETD:etd-01302008-111854
Year: 1998
DOI: 10.7907/1d4m-j065
<p>This thesis addresses several key issues in mechanics and automated planning of workpiece fixturing and robotic grasping, including accurate and efficient modelling of compliance, well-defined and practically useful quality measures, and well-defined kinematic metric functions for rigid bodies.</p>
<p>The accurate and efficient modelling of compliant fixtures and grasps is considered. A stiffness matrix formula is derived using the overlap compliance representation for quasi-rigid bodies. In contrast to existing approaches using the linear contact model, this formula is well-suited to automated planning algorithms since it can incorporate realistic nonlinear contact models (e.g., the classical Hertz model), and can be directly computed from CAD data on basic geometric and material properties of the bodies. The formula is then used as a basis for a systematic analysis of local curvature effects on fixture stability. This analysis shows that destabilizing effects of local curvatures are practically negligible, and that curvature effects can be used to stabilize, sometimes significantly, an otherwise unstable fixture. The stiffness matrix formula is also used to show that stability analysis in general depends on the choice of contact models, which offers additional evidence for the importance of using realistic contact models.</p>
<p>The stiffness and deflection quality measures are defined for compliant fixtures and grasps, and are applied to optimal planning. Unlike existing quality measures that rely on heuristic rules or depend on reference frame choices, the stiffness and deflection quality measures are theoretically sound. Equally important is that these quality measures accurately characterize functional performances which are important to practical fixturing applications, such as fixture stiffness and workpiece deflections. The stiffness and deflection quality measures are applied to optimal fixture and grasp planning, resulting in maximum-stiffness and minimum-deflection fixtures and grasps. The qualitative properties of optimal fixtures are characterized with respect to each quality measure, and efficient techniques are developed for finding such optimal fixtures.</p>
<p>The final key issue is concerned with formal well-definedness conditions and practical development methods for rigid body kinematic metric functions, such as norms, inner products, and distance metrics. Based on an intrinsic definition of the configuration space of a rigid body, the notion of objectivity is proposed to formalize the natural requirement that metric measurements be indifferent to the observers who perform the measurements. This notion is then used to clarify the fundamental physical implications of left, right and bi-invariant functions on SE(3), and is further shown to be equivalent to the notion of frame-invariance. Based on these clarifications, several frame-invariant norms of rigid body velocities and wrenches, which have interesting physical interpretations, are defined.</p>https://resolver.caltech.edu/CaltechETD:etd-01302008-111854Modeling and Experiments for a Class of Robotic Endoscopes
https://resolver.caltech.edu/CaltechETD:etd-10112006-154843
Year: 1999
DOI: 10.7907/NG6V-TD44
<p>Current developments in minimally invasive medical practice motivated this study of self-propelled, robotic endoscopes for deep penetration into curved physiological lumens. The conceptual design of such devices is applicable to endoscopy within a variety of lumens in the human body, e.g., blood vessels, but the initial objective of this technology is to provide access to the interior of the entire small intestine without surgical incisions. The small intestine presents several challenges to endoscopic penetration: it is extremely compliant to applied loading, internally lubricated, easily injured, and contains many tight curves along its length of approximately eighteen feet.</p>
<p>This thesis reports the basic design and locomotion concepts for one class of endoscopic robots that are intended to provide safe and reliable traversal of the small intestine via worm-like actuation. Five generations of proof-of-concept prototype robots have been built to validate the fundamental concepts. Furthermore, these miniaturized robots have incorporated the following features: redundant actuation with computer control, tool-free modular assembly, and on-board videoimaging capability. The prototypes have been tested in rubber tubing, the small intestines of deceased pigs, and in the small intestines of live, anaesthetized pigs.</p>
<p>At the onset of this research, little regarding the elastic properties of small intestine existed in the biomechanics literature that would be applicable to the design of these mechanisms. However, accurate prediction of the small intestine's response to robotic loadings would dramatically improve the research and development process of these machines. Thus, an investigation of the elastic behavior of the small intestine commenced. Finite deformation, nonlinear, anisotropic, incompressible, viscoelastic behavior of the small intestine was studied. This soft tissue biomechanical analysis and experimentation (on living and dissected intestinal specimens) culminated with a numerical model that simulates intestinal response to the actions of a prototypical robotic component. Experiments on living specimens were performed to determine the levels of applied loadings and internal stresses that are likely to injure these fragile tissues, and the biomechanics computer modeling incorporates three distinct measures for injury potential.</p>https://resolver.caltech.edu/CaltechETD:etd-10112006-154843Automatic observation and synthesis of human motion
https://resolver.caltech.edu/CaltechETD:etd-11022006-104150
Year: 2000
DOI: 10.7907/n3fn-jd79
Over the past few decades Computer Vision and Computer Graphics have experienced a rapid evolution, thanks in part to the continual improvement in computer hardware, which enables the investigation of increasingly complex problems.
In Computer Graphics this evolution is visible on a nearly day-by-day basis. For instance, computer-generated special effects in feature films have evolved to such a level of sophistication that it is often impossible to distinguish what is real from what is not. However, one challenging problem that still stands, considered by many experts in the field to be a Holy-Grail of Computer Graphics, is the automatic synthesis of life-like human character animation. Although rendering and modeling techniques have reached a stage where a computer generated image of a person is nearly indistinguishable from the real thing, as soon as that model begins to move the illusion is broken. The problem is difficult because no-one yet knows how to model human motion in all it's intricacy and subtlety, and also because humans are so well tuned to perceive these subtleties that they can only be fooled if the modeling is done with complete perfection.
In this thesis, we explore a novel method of automatic synthesis of human motion that brings us one step closer to the ultimate goal. The method is based on decomposing human motion into elemental, nameable actions such as walking, running, and throwing, and using observations of people performing these actions to create mathematical models of the actions. Various samples of an action are acquired, and each sample is labeled according to state (initial body configuration), goal (desired outcome of the motion, such as direction of a throw or placement of a foot for a step), and mood and style parameters. Then established and novel techniques of machine learning are applied to derive a function that can synthesis a motion given some desired parameters. We explore the use of polynomial interpolants, radial basis function networks (RBFs), feed-forward neural networks (FFNNs) with sigmoidal activation functions, as well as a new method with local linear models. We find that a linear model more often that not works quite well, whereas higher order polynomial interpolants, RBFs and FFNNs are unable to extrapolate robustly when the motion parameters lie outside of the convex hull of the parameters of the available sample motions. The method with local linear models successfully improves the fidelity of the synthetic motions compared to the linear model, and also provides robust extrapolation. We also investigate the use of a recursive, probabilitic model where motions are specified by defining the initial and final body poses of the motion, and synthesis is done by computing the most likely motion to satisfy the boundary constraints. Although the results with this method are not yet completely satisfactory, it holds promise, and under certain types of conditions can re-synthesize the sample motions more accurately than any of the other methods.
With the additional development of methods to smoothly concatenate actions together and to interactively map synthesized motions to a 3-D polygonal character model, a realtime interactive demo was created that successfully demonstrates the level of realism and interactivity achievable by our method of human motion synthesis.
Our interest in the problem of realistic human motion synthesis arose from an initial study of the (in some sense) inverse problem in Computer Vision of the automatic observation (rather than synthesis) of human motion. Although progress in Computer Vision has not yet reached a level enabling it's widespread use in daily life, this state will most likely be achieved within the next decade. One large class of problems for which this is the case is the endowment of computers with visual perceptual skills similar to those of humans. Among the vast set of visual tasks imagineable, the automatic detection, recognition, and estimation of humans and human motion is a particularly interesting set of problems since there are many possible applications of such a technology in modern life, ranging from security and monitoring systems, to systems for biometric analysis, to novel human-machine interfaces.
In this thesis we describe a method of robustly estimating the motion of a human body from a monocular view. The method is based on the use of a 3-D model of the body, and comparing the actual image to an expected image based on the 3-D model to update the estimate of the body pose at each time step. The method was implemented in realtime as a human-machine interface. This system demonstrated that the method can be used to robustly track a human arm with a hand-tip positioning resolution of 2cm under close viewing conditions (where perspective projection causes significant changes in the appearance of the arm in the camera view).
https://resolver.caltech.edu/CaltechETD:etd-11022006-104150Robust mask-layout and process synthesis in micro-electro-mechanical-systems (MEMS) using genetic algorithms
https://resolver.caltech.edu/CaltechETD:etd-08302005-131428
Year: 2001
DOI: 10.7907/e8hm-z754
This thesis reports a Genetic Algorithm approach for the mask-layout and process flow synthesis problem. For a given desired target shape, an optimal mask-layout and process flow can be automatically generated using the Genetic Algorithm synthesis approach. The Genetic Algorithm manipulates and evolves a population of candidate solutions (mask-layouts and process parameters) by utilizing a process simulation tool to evaluate the performance of the candidate solutions. For the mask-layout and process flow synthesis problem, encoding schemes, selection schemes, and genetic operations have been developed to effectively explore the solution space and control the evolution and convergence of the solutions.
The synthesis approach is tested for mask-layout and process synthesis for bulk wet etching. By integrating a bulk wet etching simulation tool into the Genetic Algorithm iterations, the algorithm can automatically generate proper mask-layout and process flow which can fabricate 3-D geometry close to the desired 3-D target shape. For structures with convex corners, complex compensation structures can be synthesized by the algorithm. More importantly, the process flow can also be synthesized. For multi-step wet etching processes, proper etchant sequence and etch times for each etch step can be synthesized automatically by the algorithm. When the choice of different process flows exists, the enlarged solution space makes the design problem more challenging. The ability to synthesize process flows makes the automatic design method more complete and more valuable.
The algorithm is further extended to achieve robust design. Since fabrication variations and modeling inaccuracy always exist, the synthesized solutions without considering these variations may not generate satisfactory results in actual fabrication. Robust design methods are developed to synthesize robust mask-layouts and process flows in "noisy" environment. Since the synthesis procedure considers the effect of variations in the fabrication procedures, the final synthesized solution will have high robustness to the variations, and will generate satisfactory results under a variety of fabrication conditions. The robust design approaches are implemented and tested for robust mask-layout design for mask misalignment and etch rate variations. Mask-layouts robust to mask misalignment noise and etch rate variations during the fabrication can be synthesized. The synthesized mask-layouts generally improve the yield significantly by exhibiting consistent performance under a variety of fabrication conditions.https://resolver.caltech.edu/CaltechETD:etd-08302005-131428Parylene MEMS Technology for Adaptive Flow Control of Flapping Flight
https://resolver.caltech.edu/CaltechETD:etd-09162005-110430
Year: 2002
DOI: 10.7907/jtjd-dd33
This thesis is the culmination of research work in developing a parylene MEMS technology to fabricate MEMS wings and large-area parylene actuator skins for real-time adaptive flow control for flapping flight applications.
In this thesis, the novel MEMS-based wing technology is presented using titanium-alloy metal (Ti-6A1-4V) as wingframe and parylene-C as wing membrane. With this technology, the ability to produce light, yet robust, 3-D wings can be achieved. The use of MEMS technology enables systematic research in terms of repeatability, size control, weight minimization, and mass production of the wings. By fabricating the wing with the photolithography and etching techniques, fast turnaround time of various wing designs can be easily obtained. The wings are optimized to utilize the flow separation to achieve a high lift coefficient, C(L), as large as five times that of the fixed-wing aircraft. The aerodynamic tests are performed in a high quality low-speed wind tunnel with velocity uniformity of 0.5% and speeds range from 1 to 10 m/s. The wind-tunnel test results are presented and discussed.
As part of the investigation to integrate MEMS actuators onto the wings for realtime adaptive flow control, the MEMS technology is developed to fabricate the first large-area wafer-sized, flexible parylene MEMS electrostatic actuator skins. The technology is first developed to fabricate parylene actuator diaphragm on a silicon chip. The actuator diaphragm is made of two metallized layers of parylene membranes with offset vent holes. Without electrostatic actuation, air can move freely from one side of the skin to the other side through the vent holes. With actuation, these vent holes are sealed and the airflow is controlled. The membrane behaves as a complete diaphragm. This function is successfully demonstrated using a 2-mm x 2-mm parylene diaphragm electrostatic actuator valves.
Finally, this technology is applied to fabricate large area wafer-sized actuator skins. The skins contain only parylene and metalized electrodes and have no bulk silicon as a structural component. Plate and check-valved skin types are fabricated and both are integrated onto the MEMS wings for aerodynamic flow control. The integration of micro-valved actuators has shown significant effect on the aerodynamic performance of the flapping flight. The wind-tunnel test results are analyzed and discussed in detail in this thesis.
https://resolver.caltech.edu/CaltechETD:etd-09162005-110430Optimal and Cooperative Control of Vehicle Formations
https://resolver.caltech.edu/CaltechETD:etd-10242005-105000
Year: 2002
DOI: 10.7907/M4N7-AK02
Control of vehicle formations has emerged as a topic of significant interest to the controls community. In applications such as microsatellites and underwater vehicles, formations have the potential for greater functionality and versatility than individual vehicles. In this thesis, we investigate two topics relevant to control of vehicle formations: optimal vehicle control and cooperative control.
The framework of optimal control is often employed to generate vehicle trajectories. We use tools from geometric mechanics to specialize the two classical approaches to optimal control, namely the calculus of variations and the Hamilton-Jacobi-Bellman (HJB) equation, to the case of vehicle dynamics. We employ the formalism of the covariant derivative, useful in geometric representations of vehicle dynamics, to relate variations of position to variations of velocity. When variations are computed in this setting, the evolution of the adjoint variables is shown to be governed by the covariant derivative, thus inheriting the geometric structure of the vehicle dynamics. To simplify the HJB equation, we develop the concept of time scalability enjoyed by many vehicle systems. We employ this property to eliminate time from the HJB equation, yielding a purely spatial PDE whose solution supplies both finite-time optimal trajectories and a time-invariant stabilizing control law.
Cooperation among vehicles in formation depends on intervehicle communication. However, vehicle communication is often subject to disruption, especially in an adversarial setting. We apply tools from graph theory to relate the topology of the communication network to formation stability. We prove a Nyquist criterionthat uses the eigenvalues of the graph Laplacian matrix to determine the effect of the graph on formation stability. We also propose a method for decentralized information exchange between vehicles. This approach realizes a dynamical system that supplies each vehicle with a common reference to be used for cooperative motion. We prove a separation principle that states that formation stability is achieved if the information flow is stable for the given graph and if the local controller stabilizes the vehicle. The information flow can be rendered highly robust to changes in the graph, thus enabling tight formation control despite limitations in intervehicle communication capability.https://resolver.caltech.edu/CaltechETD:etd-10242005-105000Optimal and Cooperative Control of Vehicle Formations
https://resolver.caltech.edu/CaltechETD:etd-10242005-105000
Year: 2002
DOI: 10.7907/M4N7-AK02
Control of vehicle formations has emerged as a topic of significant interest to the controls community. In applications such as microsatellites and underwater vehicles, formations have the potential for greater functionality and versatility than individual vehicles. In this thesis, we investigate two topics relevant to control of vehicle formations: optimal vehicle control and cooperative control.
The framework of optimal control is often employed to generate vehicle trajectories. We use tools from geometric mechanics to specialize the two classical approaches to optimal control, namely the calculus of variations and the Hamilton-Jacobi-Bellman (HJB) equation, to the case of vehicle dynamics. We employ the formalism of the covariant derivative, useful in geometric representations of vehicle dynamics, to relate variations of position to variations of velocity. When variations are computed in this setting, the evolution of the adjoint variables is shown to be governed by the covariant derivative, thus inheriting the geometric structure of the vehicle dynamics. To simplify the HJB equation, we develop the concept of time scalability enjoyed by many vehicle systems. We employ this property to eliminate time from the HJB equation, yielding a purely spatial PDE whose solution supplies both finite-time optimal trajectories and a time-invariant stabilizing control law.
Cooperation among vehicles in formation depends on intervehicle communication. However, vehicle communication is often subject to disruption, especially in an adversarial setting. We apply tools from graph theory to relate the topology of the communication network to formation stability. We prove a Nyquist criterionthat uses the eigenvalues of the graph Laplacian matrix to determine the effect of the graph on formation stability. We also propose a method for decentralized information exchange between vehicles. This approach realizes a dynamical system that supplies each vehicle with a common reference to be used for cooperative motion. We prove a separation principle that states that formation stability is achieved if the information flow is stable for the given graph and if the local controller stabilizes the vehicle. The information flow can be rendered highly robust to changes in the graph, thus enabling tight formation control despite limitations in intervehicle communication capability.https://resolver.caltech.edu/CaltechETD:etd-10242005-105000Parylene MEMS Technology for Adaptive Flow Control of Flapping Flight
https://resolver.caltech.edu/CaltechETD:etd-09162005-110430
Year: 2002
DOI: 10.7907/jtjd-dd33
This thesis is the culmination of research work in developing a parylene MEMS technology to fabricate MEMS wings and large-area parylene actuator skins for real-time adaptive flow control for flapping flight applications.
In this thesis, the novel MEMS-based wing technology is presented using titanium-alloy metal (Ti-6A1-4V) as wingframe and parylene-C as wing membrane. With this technology, the ability to produce light, yet robust, 3-D wings can be achieved. The use of MEMS technology enables systematic research in terms of repeatability, size control, weight minimization, and mass production of the wings. By fabricating the wing with the photolithography and etching techniques, fast turnaround time of various wing designs can be easily obtained. The wings are optimized to utilize the flow separation to achieve a high lift coefficient, C(L), as large as five times that of the fixed-wing aircraft. The aerodynamic tests are performed in a high quality low-speed wind tunnel with velocity uniformity of 0.5% and speeds range from 1 to 10 m/s. The wind-tunnel test results are presented and discussed.
As part of the investigation to integrate MEMS actuators onto the wings for realtime adaptive flow control, the MEMS technology is developed to fabricate the first large-area wafer-sized, flexible parylene MEMS electrostatic actuator skins. The technology is first developed to fabricate parylene actuator diaphragm on a silicon chip. The actuator diaphragm is made of two metallized layers of parylene membranes with offset vent holes. Without electrostatic actuation, air can move freely from one side of the skin to the other side through the vent holes. With actuation, these vent holes are sealed and the airflow is controlled. The membrane behaves as a complete diaphragm. This function is successfully demonstrated using a 2-mm x 2-mm parylene diaphragm electrostatic actuator valves.
Finally, this technology is applied to fabricate large area wafer-sized actuator skins. The skins contain only parylene and metalized electrodes and have no bulk silicon as a structural component. Plate and check-valved skin types are fabricated and both are integrated onto the MEMS wings for aerodynamic flow control. The integration of micro-valved actuators has shown significant effect on the aerodynamic performance of the flapping flight. The wind-tunnel test results are analyzed and discussed in detail in this thesis.
https://resolver.caltech.edu/CaltechETD:etd-09162005-110430Model Selection, Identification and Robust Control for Dynamical Systems
https://resolver.caltech.edu/CaltechTHESIS:02082012-113016272
Year: 2002
DOI: 10.7907/YEV9-8X44
<p>To fully exploit new technologies for response mitigation and structural health monitoring, improved system identification and controller design methodologies are desirable that explicitly treat all the inherent uncertainties. In this thesis, a probabilistic framework is presented for model selection, identification and robust control of smart structural systems under dynamical loads, such as those induced by wind or earthquakes. First, a probabilistic based approach is introduced for selecting the most plausible class of models for a dynamical system using its response measurements. The proposed approach allows for quantitatively comparing the plausibility of different classes of models among a specified set of classes.</p>
<p>Then, two probabilistic identification techniques are presented. The first one is for modal identification using nonstationary response measurements and the second one is for updating nonlinear models using incomplete noisy measurements only. These methods allow for updating of the uncertainties associated with the values of the parameters controlling the dynamic behavior of the structure by using noisy response measurements only. The probabilistic framework is very well-suited for solving this nonunique problem and the updated probabilistic description of the system can be used to design a robust controller of the system. It can also be used for structural health monitoring.</p>
<p>Finally, a reliability-based stochastic robust control approach is used to design the controller for an active control system. Feedback of the incomplete response at earlier time steps is used, without any state estimation. The optimal controller is chosen by minimizing the robust failure probability over a set of possible models for the system. Here, failure means excessive levels of one or more response quantities representative of the performance of the structure and the control devices. When calculating the robust failure probability, the plausibility of each model as a representation of the system's dynamic behavior is quantified by a probability distribution over the set of possible models; this distribution is initially based on engineering judgement, but it can be updated using the aforementioned system identification approaches if dynamic data become available from the structure. Examples are presented to illustrate the proposed controller design procedure, which includes the procedure of model selection, identification and robust control for smart structures.</p> https://resolver.caltech.edu/CaltechTHESIS:02082012-113016272Methods for the Analysis of Visual Motion
https://resolver.caltech.edu/CaltechTHESIS:10072010-140426638
Year: 2002
DOI: 10.7907/SAVS-VJ68
Vision is a primary sense that allows human beings to interact with their environment and motion is one of the most important cues that vision can explore and utilize.
In this thesis, we present computational approaches to the problems of inferring three-dimensional motion information and perceiving two-dimensional human motions from a sequence of images captured by a camera.
The three-dimensional structure of world can be represented by distinguishable features, such as points. Assume all the features move under the same rigid motion in space, this motion can be recovered from the projections of the features in three views by solving a set of trilinear constraints. The trilinear constraints have been considered only as algebraic equations so that their satisfactory performance in motion estimation is not easy to understand. This thesis solves this puzzle by discovering a geometrical interpretation of trilinear constraints. It is showed that those algebraic equations correspond to depth errors appropriately weighted by a function of the relative reliability of the corresponding measurements. When the assumption is relaxed to allowing features to move under different rigid motions, this thesis proposes a three-dimensional motion based expectation-maximization algorithm combined with the modified separation matrix scheme to cluster the features undergoing the same motion into a group and estimate the motion for every group at the same time.
The problem of detecting and recognizing human motions arises from many applications in computer vision. This thesis describes an algorithm to detect human body from their motion patterns in a pair of frames which is based on learning an approximate probabilistic model of the positions and velocities of body joints. It then presents a scheme to recognize human actions in a sequence of frames assuming the human body is detected. This scheme enables us to simultaneously recognize both the action and the body poses in the observed sequence.
All our theoretical work is supported by experimental results.
https://resolver.caltech.edu/CaltechTHESIS:10072010-140426638Functional Magnetic Resonance Imaging in Rhesus Macaque Monkeys
https://resolver.caltech.edu/CaltechTHESIS:01252012-103337039
Year: 2002
DOI: 10.7907/JBRJ-AS69
This thesis presents a method for functional magnetic resonance imaging in the brain of the rhesus macaque monkey, Macaca muialla. Experiments were performed in an awake-behaving animal at 1.5 T in a conventional clinical magnetic resonance system. Strategies to train the animal within the MR environment and to ensure behavioral compliance are described. Limitations of studying macaque functional neuroanatomy at this magnetic field strength are also discussed. Methods to improve signal-to-noise beyond the scope of conventional BOLD imaging at 1.5 T are presented, including the use of very high magnetic fields for functional imaging (11.7 T), as well as novel methods to improve sensitivity using intravascular iron oxide contrast media. Using the techniques developed for this thesis, a series of studies are presented to examine the visual pathways in the primate brain, allowing direct comparison of functional neuroanatomy between nonhuman primates and human cortex. Although the two species are anatomically different, direct functional homology within the visual cortex is demonstrated.
https://resolver.caltech.edu/CaltechTHESIS:01252012-103337039Self-Organized Robotic System Design and Autonomous Odor Localization
https://resolver.caltech.edu/CaltechETD:etd-06112002-132034
Year: 2002
DOI: 10.7907/KH16-XH33
This thesis presents a methodology for designing self-organized autonomous robotic systems and demonstrates how this process can be applied to the problem of finding the source of an airborne odor plume. The design methodology is applicable to other task domains and the resulting odor localization system extends the state of the art.
The design procedure centers on the ability to define a specific task performance metric, systematically evaluate performance in a realistic environment, and define abstract relationships between system parameters and system performance. Once such relationships have been experimentally validated in a test environment, they can be used to guide the design of a deployable system. Because this process relies heavily on evaluative feedback, this work emphasizes the development of tools that allow the collection of accurate performance data. It presents a reliable multiple robot test-bed and some task-enabling sensory hardware, as well as validation of the sensory and kinematic models used in simulation. Also, a reinforcement learning methodology is described that provides consistent optimization performance while minimizing the amount of required evaluation.
The design methodology is applied to the task of odor localization. Specifically, this thesis analyzes a basic collective search task and derives the optimal group size and expected performance bounds for random and coordinated search. It also investigates a set of biologically inspired behaviors that permit an agent to traverse an odor plume to its source and describes the common characteristics of successful algorithms. One of these algorithms is implemented on the real test-bed and in simulation to verify that plume traversal is taking place and that the use of multiple collaborating robots can expand the reachable performance space. Collective search and plume traversal are then combined (along with egocentric source declaration) into the full odor localization task which is optimized in simulation. Then, following the design methodology, a model is presented which can aid in the prediction of performance and choice of algorithm parameters in more complex environments. Finally, a flocking behavior is designed, and the addition of this flocking behavior to the plume tracing algorithm is shown to produce a more capable systemhttps://resolver.caltech.edu/CaltechETD:etd-06112002-132034Control of Multiple Model Systems
https://resolver.caltech.edu/CaltechETD:etd-07312002-091923
Year: 2002
DOI: 10.7907/17Q7-Y019
This thesis considers the control of multiple model systems. These are systems for which only one model out of some finite set of models gives the system dynamics at any given time. In particular, the model that gives the system dynamics can change over time. This thesis covers some of the theoretical aspects of these systems, including controllability and stabilizability. As an application, ``overconstrained' mechanical systems are modeled as multiple model systems. Examples of such systems include distributed manipulation problems such as microelectromechanical systems and many wheeled vehicles such as the Sojourner vehicle of the Mars Pathfinder mission. Such systems are typified by having more Pfaffian constraints than degrees of freedom. Conventional classical motion planning and control theories do not directly apply to overconstrained systems. Control issues for two examples are specifically addressed. The first example is distributed manipulation. Distributed manipulation systems control an object's motion through contact with a high number of actuators. Stability results are shown for such systems and control schemes based on these results are implemented on a distributed manipulation test-bed. The second example is that of overconstrained vehicles, of which the Mars rover is an example. The nonlinear controllability test for multiple model systems is used to answer whether a kinematic model of the rover is or is not controllable.https://resolver.caltech.edu/CaltechETD:etd-07312002-091923Controlled Lagrangian and Hamiltonian Systems
https://resolver.caltech.edu/CaltechTHESIS:10112010-161816245
Year: 2002
DOI: 10.7907/3DRR-ZV53
<p>any control systems are mechanical systems. The unique feature of mechanical systems is the notion of energy, which gives much information on the stability of equilibria. Two kinds of forces are associated with the energy: dissipative force and gyroscopic force. A dissipative force is, by definition, a force which decreases the energy, and a gyroscopic force is, by definition, a force that does not change the energy. Gyroscopic forces add couplings to the dynamics. In this thesis, we develop a control design methodology which makes full use of these three physical notions: energy, dissipation, and coupling.</p>
<p>First, we develop the method of controlled Lagrangian systems. It is a systematic procedure for designing stabilizing controllers for mechanical systems by making use of energy, dissipative forces, and gyroscopic forces. The basic idea is as follows: Suppose that we are given a mechanical system and want to design a controller to asymptotically stabilize an equilibrium of interest. We look for a feedback control law such that the closed-loop dynamics can be also described by a new Lagrangian with a dissipative force and a gyroscopic force where the energy of the new Lagrangian has a minimum at the equilibrium. Then we check for asymptotic stability by applying the Lyapunov stability theory with the new energy as a Lyapunov function.</p>
<p>Next, we show that the method of controlled Lagrangian systems and its Hamiltonian counterpart, the method of controlled Hamiltonian systems, are equivalent for simple mechanical systems where the underlying Lagrangian is of the form kinetic minus potential energy. In addition, we extend both the Lagrangian and Hamiltonian sides of this theory to include systems with symmetry and discuss the relevant reduction theory.</p>https://resolver.caltech.edu/CaltechTHESIS:10112010-161816245Controlled Lagrangian and Hamiltonian Systems
https://resolver.caltech.edu/CaltechTHESIS:10112010-161816245
Year: 2002
DOI: 10.7907/3DRR-ZV53
<p>any control systems are mechanical systems. The unique feature of mechanical systems is the notion of energy, which gives much information on the stability of equilibria. Two kinds of forces are associated with the energy: dissipative force and gyroscopic force. A dissipative force is, by definition, a force which decreases the energy, and a gyroscopic force is, by definition, a force that does not change the energy. Gyroscopic forces add couplings to the dynamics. In this thesis, we develop a control design methodology which makes full use of these three physical notions: energy, dissipation, and coupling.</p>
<p>First, we develop the method of controlled Lagrangian systems. It is a systematic procedure for designing stabilizing controllers for mechanical systems by making use of energy, dissipative forces, and gyroscopic forces. The basic idea is as follows: Suppose that we are given a mechanical system and want to design a controller to asymptotically stabilize an equilibrium of interest. We look for a feedback control law such that the closed-loop dynamics can be also described by a new Lagrangian with a dissipative force and a gyroscopic force where the energy of the new Lagrangian has a minimum at the equilibrium. Then we check for asymptotic stability by applying the Lyapunov stability theory with the new energy as a Lyapunov function.</p>
<p>Next, we show that the method of controlled Lagrangian systems and its Hamiltonian counterpart, the method of controlled Hamiltonian systems, are equivalent for simple mechanical systems where the underlying Lagrangian is of the form kinetic minus potential energy. In addition, we extend both the Lagrangian and Hamiltonian sides of this theory to include systems with symmetry and discuss the relevant reduction theory.</p>https://resolver.caltech.edu/CaltechTHESIS:10112010-161816245Model Selection, Identification and Robust Control for Dynamical Systems
https://resolver.caltech.edu/CaltechTHESIS:02082012-113016272
Year: 2002
DOI: 10.7907/YEV9-8X44
<p>To fully exploit new technologies for response mitigation and structural health monitoring, improved system identification and controller design methodologies are desirable that explicitly treat all the inherent uncertainties. In this thesis, a probabilistic framework is presented for model selection, identification and robust control of smart structural systems under dynamical loads, such as those induced by wind or earthquakes. First, a probabilistic based approach is introduced for selecting the most plausible class of models for a dynamical system using its response measurements. The proposed approach allows for quantitatively comparing the plausibility of different classes of models among a specified set of classes.</p>
<p>Then, two probabilistic identification techniques are presented. The first one is for modal identification using nonstationary response measurements and the second one is for updating nonlinear models using incomplete noisy measurements only. These methods allow for updating of the uncertainties associated with the values of the parameters controlling the dynamic behavior of the structure by using noisy response measurements only. The probabilistic framework is very well-suited for solving this nonunique problem and the updated probabilistic description of the system can be used to design a robust controller of the system. It can also be used for structural health monitoring.</p>
<p>Finally, a reliability-based stochastic robust control approach is used to design the controller for an active control system. Feedback of the incomplete response at earlier time steps is used, without any state estimation. The optimal controller is chosen by minimizing the robust failure probability over a set of possible models for the system. Here, failure means excessive levels of one or more response quantities representative of the performance of the structure and the control devices. When calculating the robust failure probability, the plausibility of each model as a representation of the system's dynamic behavior is quantified by a probability distribution over the set of possible models; this distribution is initially based on engineering judgement, but it can be updated using the aforementioned system identification approaches if dynamic data become available from the structure. Examples are presented to illustrate the proposed controller design procedure, which includes the procedure of model selection, identification and robust control for smart structures.</p> https://resolver.caltech.edu/CaltechTHESIS:02082012-113016272Averaging and Control of Nonlinear Systems
https://resolver.caltech.edu/CaltechETD:etd-05282003-094253
Year: 2003
DOI: 10.7907/N7HH-PM67
<p>This dissertation investigates three principal areas regarding the dynamics and control of nonlinear systems: averaging theory, controllability of mechanical systems, and control of underactuated nonlinear systems. The most effective stabilizing controllers for underactuated nonlinear systems are time-periodic, which leads to the study of averaging theory for understanding the nonlinear effect generated by resonant oscillatory inputs.</p>
<p>The research on averaging theory generalizes averaging theory to arbitrary order by synthesizing series expansion methods for nonlinear time-varying vector fields and their flows with nonlinear Floquet theory. It is shown that classical averaging theory is the application of perturbation methods in conjunction with nonlinear Floquet theory. Many known properties and consequences of averaging theory are placed within a single framework.</p>
<p>The generalized averaging theory is merged with controllability analysis of underactuated nonlinear systems to derive exponentially stabilizing controllers. Although small-time local controllability (STLC) is easily demonstrated for driftless systems via the Lie algebra rank condition, STLC for systems with drift is more complicated. Furthermore, there exists a variety of techniques and canonical forms for determining STLC. This thesis exploits notions of geometric homogeneity to show that STLC results for a large class of mechanical systems with drift can be recovered by considering a class of nonlinear dynamical systems satisfying certain homogeneity conditions. These theorems generalize the controllability results for simple mechanical control systems found in Lewis and Murray [85]. Most nonlinear controllability results for classes of mechanical systems may be obtained using these methods.</p>
<p>The stabilizing controllers derived using the generalized averaging theory and STLC analysis can be used to stabilize both systems with and without drift. Furthermore, they result in a set of tunable gains and oscillatory parameters for modification and improvement of the feedback strategy. The procedure can not only derive known controllers from the literature, but can also be used to improve them. Examples demonstrate the diversity of controllers constructed using the generalized averaging theory.</p>
<p>This dissertation concludes with a chapter devoted to biomimetic and biomechanical locomotive control systems that have been stabilized using the generalized averaging theory and the controller construction procedure. The locomotive control systems roll, wriggle, swim, and walk, demonstrating the universal nature of the control strategy proposed.</p>https://resolver.caltech.edu/CaltechETD:etd-05282003-094253MEMS Technology and Devices for a Micro Fluid Dosing System
https://resolver.caltech.edu/CaltechETD:etd-05182003-163704
Year: 2003
DOI: 10.7907/MRDS-FC06
Microelectromechanical systems (MEMS) technology has matured to the point where practical biological and chemical applications are possible. One particularly active research area is in the development of lab-on-a-chip type systems. In order to create successful lab-on-a-chip and other microfluidic systems, it is necessary to have the capability of controlling and directing fluid flow. Such functionality can be found on the front end of a microfluidic system and is known as a fluid delivery or dosing subsystem. For a MEMS micro fluid dosing system to be realized, several components are necessary. The essential components include a fluid actuator, a fluidic control device, and micro plumbing. A prototype fluid delivery system is demonstrated here using a micropump as the fluid actuator, a thermal flow sensor as the fluidic control device, and micromachined couplers as plumbing. The technology to build these components has been developed and each of these components have been fabricated and tested. A prototype constructed of discrete components has also been demonstrated. A truly integrated, channel-based fluid dosing system can be achieved through device scaling.https://resolver.caltech.edu/CaltechETD:etd-05182003-163704Real-Time Optimal Trajectory Generation for Constrained Dynamical Systems
https://resolver.caltech.edu/CaltechETD:etd-06022003-114340
Year: 2003
DOI: 10.7907/1X68-E370
<p>With the advent of powerful computing and efficient computational algorithms, real-time solutions to constrained optimal control problems are nearing a reality. In this thesis, we develop a computationally efficient Nonlinear Trajectory Generation (NTG) algorithm and describe its software implementation to solve, in real-time, nonlinear optimal trajectory generation problems for constrained systems. NTG is a nonlinear trajectory generation software package that combines nonlinear control theory, B-spline basis functions, and nonlinear programming. We compare NTG with other numerical optimal control problem solution techniques, such as direct collocation, shooting, adjoints, and differential inclusions.</p>
<p>We demonstrate the performance of NTG on the Caltech Ducted Fan testbed. Aggressive, constrained optimal control problems are solved in real-time for hover-to-hover, forward flight, and terrain avoidance test cases. Real-time trajectory generation results are shown for both the two-degree of freedom and receding horizon control designs. Further experimental demonstration is provided with the station-keeping, reconfiguration, and deconfiguration of micro-satellite formation with complex nonlinear constraints. Successful application of NTG in these cases demonstrates reliable real-time trajectory generation, even for highly nonlinear and non-convex systems. The results are among the first to apply receding horizon control techniques for agile flight in an experimental setting, using representative dynamics and computation.</p>https://resolver.caltech.edu/CaltechETD:etd-06022003-114340Fluid Locomotion and Trajectory Planning for Shape-Changing Robots
https://resolver.caltech.edu/CaltechETD:etd-05292003-160843
Year: 2003
DOI: 10.7907/MFM1-0866
Motivated by considerations of shape changing propulsion of underwater robotic vehicles, I analyze the mechanics of deformable bodies operating in an ideal fluid. I give particular attention to fishlike robots which may be considered as one or more flexing or oscillating hydrofoils. I then describe methods of planning trajectories for a fishlike robot or any other sort of robot whose locomotion has a periodic or quasi-periodic nature.https://resolver.caltech.edu/CaltechETD:etd-05292003-160843Symmetry, Reduction and Swimming in a Perfect Fluid
https://resolver.caltech.edu/CaltechETD:etd-06042003-181857
Year: 2003
DOI: 10.7907/CE65-XM80
This thesis presents a geometric picture of a deformable body in a perfect fluid and a way to approximate its dynamics and the motion, resulting from cyclic shape deformations, of the body and, interestingly, the fluid as well. Emphasis is placed on the group structure of the configuration space of the body fluid system and the resulting symmetry in their equations of motion. Symmetry is also used to reduce a series expansion for the flow of a time dependent vector field in order to obtain a novel expansion for the path-ordered exponential. This can be used to approximate holonomy, or geometric phase, in a principal bundle when its evolution is governed by a connection on the bundle and it is subject to periodic shape inputs. Simple models for swimming in and the stirring of a perfect fluid are proposed and examined.https://resolver.caltech.edu/CaltechETD:etd-06042003-181857Spike Train Characterization and Decoding for Neural Prosthetic Devices
https://resolver.caltech.edu/CaltechETD:etd-07232003-012018
Year: 2004
DOI: 10.7907/GK20-5W75
<p>Neural prosthetic device has the potential of benefiting millions of lock-in and spinal cord injury survivors. One branch of the ongoing research is to construct reach movement based prosthetic devices. This thesis proposes statistical methods based on applying the Haar wavelet packets to spike trains in order to answer some of the questions in this field.</p>
<p>Although spike train is the most frequently used data in the neural science community, its stochastic properties are not fully understood or characterized. This thesis suggests a formal spike train characterization method using the Haar wavelet packet. Because of the multi-scale property of the wavelet packet, Poisson characteristics at different scales can be assessed. Moreover, Poisson Scale-gram is proposed to help visualize the characteristics of the spike train at different scales.</p>
<p>Because some neurons display non-Poisson characteristics, it is necessary to extract the relevant features from spike trains in the context of decoding. The thesis suggests a feature extraction method that searches all the wavelet packet coefficients for the ones with the largest discriminability, quantified by mutual information. This technique returns the most informative feature(s) in the context of the Bayesian classifier. Decoding performance of this proposed method is compared against the one using mean firing rate only on both surrogate data and the actual data from PRR.</p>
<p>It is also crucial to decode cognitive states because they provide the extra control signals necessary for practical implementation of the prosthetic devices. This thesis proposes a simple finite state machine approach along with an interpreter that interprets the decoding results and to regulate when the transition should occur. It demonstrates that the finite state machine framework, when coupled with the interpreter, offers a simple autonomous control scheme for the neuron prosthetic system envisioned.</p>
<p>While the neural prosthetic system is in its infancy, many theoretical and experimental works lay the foundation for a bright future in this field. This thesis answers the spike train characterization and decoding questions in a theoretical manner while offering several novel techniques that bring new ideas and insights into the research field.</p>https://resolver.caltech.edu/CaltechETD:etd-07232003-012018Vortex Formation and Drag on Low Aspect Ratio, Normal Flat Plates
https://resolver.caltech.edu/CaltechETD:etd-05292004-183807
Year: 2004
DOI: 10.7907/907K-2F28
<p>Experiments were done to investigate the role of vortex formation in the drag force generation of low aspect ratio, normal flat plates starting from rest. This very simplified case is a first, fundamental step toward understanding the more complicated flow of hovering flight, which relies primarily on drag for propulsion. The relative importance of the plate's free end, or tip, with varying aspect ratio was also studied.</p>
<p>Identifying the relationship among aspect ratio, vortex formation, and drag force can provide insight into the wing aspect ratios and kinematics found nature, with the eventual goal of designing man-made flapping wing micro air vehicles.</p>
<p>The experiments were carried out using flat plate models in a towing tank at a moderate Reynolds number of 3000. Two aspect ratios, 6 and 2, were considered, the latter in order to have a highly tip-dominated case. A force balance measured the time-varying drag, and multiple, perpendicular sections of the flow velocity were measured quantitatively using digital particle image velocimetry. Vorticity fields were calculated from the velocity data, and features in the drag force for different aspect ratios were related to the vortex dynamics. Finally, since the flow is highly three-dimensional, dye flow visualization was done to characterize its structure and to augment the two-dimensional digital particle image velocimetry data.</p>https://resolver.caltech.edu/CaltechETD:etd-05292004-183807Vortex Formation and Drag on Low Aspect Ratio, Normal Flat Plates
https://resolver.caltech.edu/CaltechETD:etd-05292004-183807
Year: 2004
DOI: 10.7907/907K-2F28
<p>Experiments were done to investigate the role of vortex formation in the drag force generation of low aspect ratio, normal flat plates starting from rest. This very simplified case is a first, fundamental step toward understanding the more complicated flow of hovering flight, which relies primarily on drag for propulsion. The relative importance of the plate's free end, or tip, with varying aspect ratio was also studied.</p>
<p>Identifying the relationship among aspect ratio, vortex formation, and drag force can provide insight into the wing aspect ratios and kinematics found nature, with the eventual goal of designing man-made flapping wing micro air vehicles.</p>
<p>The experiments were carried out using flat plate models in a towing tank at a moderate Reynolds number of 3000. Two aspect ratios, 6 and 2, were considered, the latter in order to have a highly tip-dominated case. A force balance measured the time-varying drag, and multiple, perpendicular sections of the flow velocity were measured quantitatively using digital particle image velocimetry. Vorticity fields were calculated from the velocity data, and features in the drag force for different aspect ratios were related to the vortex dynamics. Finally, since the flow is highly three-dimensional, dye flow visualization was done to characterize its structure and to augment the two-dimensional digital particle image velocimetry data.</p>https://resolver.caltech.edu/CaltechETD:etd-05292004-183807Variational Integrators
https://resolver.caltech.edu/CaltechETD:etd-06072004-161416
Year: 2004
DOI: 10.7907/CMZ1-RQ16
<p>Variational integrators are a class of discretizations for mechanical systems which are derived by discretizing Hamilton's principle of stationary action. They are applicable to both ordinary and partial differential equations, and to both conservative and forced problems. In the absence of forcing they conserve (multi-)symplectic structures, momenta arising from symmetries, and energy up to a bounded error.</p>
<p>In this thesis the basic theory of discrete variational mechanics for ordinary differential equations is developed in depth, and is used as the basis for constructing variational integrators and analyzing their numerical properties. This is then taken as the starting point for the development of a new class of asynchronous time stepping methods for solid mechanics, known as Asynchronous Variational Integrators (AVIs). These explicit methods time step different elements in a finite element mesh with fully independent and decoupled time steps, allowing the simulation to proceed locally at the fastest rate allowed by local stability restrictions. Numerical examples of AVIs are provided, demonstrating the excellent properties they posess by virtue of their variational derivation.</p>https://resolver.caltech.edu/CaltechETD:etd-06072004-161416Nanowicking: Multi-scale Flow Interaction with Nanofabric Structures
https://resolver.caltech.edu/CaltechETD:etd-04202005-172426
Year: 2005
DOI: 10.7907/XMW0-DJ25
<p>Dense arrays of aligned carbon nanotubes are designed into strips — nanowicks — as a miniature wicking element for liquid delivery and potential microfluidic chemical analysis devices. The delivery function of nanowicks enables novel fluid transport devices to run without any power input, moving parts or external pump. The intrinsically nanofibrous structure of nanowicks provides a sieving matrix for molecular separations, and a high surface-to-volume ratio porous bed to carry catalysts or reactive agents.</p>
<p>This work also experimentally studies the spontaneous fluid transport along nanowicks. Liquid is conveyed through corner flow, surface flow, and interstitial flow through capillary force and the Marangoni effect. The main course for corner flow and surface flow follows Washburn behavior, and can deliver liquid centimeters away from the input blob with a speed on the order of millimeters per second depending on the nanowick configuration and the amount of input liquid. Corner flow can be minimized and even eliminated through proper nanowick and input design. Otherwise, corner flow interacts with surface flow in the first 2mm of the pathway closest to the input point. Interstitial flow dominates the late stage. It is driven by both capillary force and concentration-gradient-induced Marangoni force. The concentration gradient is determined by two competing rates: surfactant diffusion in solution and adsorption onto nanotube surfaces. The flow inside nanowicks may wick hundreds of microns in seconds or tens of seconds. A non-conventional advancing front may develop in the flow around nanowicks. They are seen as (i) Rayleigh instability-induced fingering in surface flow on millimeter-wide nanowicks, (ii) viscous instability-induced branching near almost-stagnant surface film at low surfactant concentration, and (iii) disjointed wetting domains at very low concentration.</p>https://resolver.caltech.edu/CaltechETD:etd-04202005-172426Unsteady Fluid Mechanics of Starting-Flow Vortex Generators with Time-Dependent Boundary Conditions
https://resolver.caltech.edu/CaltechETD:etd-04112005-151435
Year: 2005
DOI: 10.7907/QV8Y-YZ12
<p>Nature has repeatedly converged on the use of starting flows for mass, momentum, and energy transport. The vortex loops that form during flow initiation have been reproduced in the laboratory and have been shown to make a proportionally larger contribution to fluid transport than an equivalent steady jet. However, physical processes limit growth of the vortex loops, suggesting that these flows may be amenable to optimization. Although it has been speculated that optimal vortex formation might occur naturally in biological systems, previous efforts to quantify the biological mechanisms of vortex formation have been inconclusive. In addition, the unsteady fluid dynamical effects associated with starting flow vortex generators are poorly understood.</p>
<p>This thesis describes a combination of new experimental techniques and in vivo animal measurements that determine the effects of fluid-structure interactions on vortex formation by starting flow propulsors. Results indicate that vortex formation across various biological systems is manipulated by these kinematics in order to maximize thrust and/or propulsive efficiency. An emphasis on observed vortex dynamics and transient boundary conditions facilitates quantitative comparisons across fluid transport schemes, irrespective of their individual biological functions and physical scales.</p>
<p>The primary contributions of this thesis are the achievement of quantitative measures of unsteady vortex dynamics via fluid entrainment and added-mass effects, and the development of a robust framework to facilitate the discovery of general design principles for effective fluid transport in engineering technologies and biological therapies. The utility of this new research paradigm is demonstrated through a variety of examples.</p>https://resolver.caltech.edu/CaltechETD:etd-04112005-151435Nanowicking: Multi-scale Flow Interaction with Nanofabric Structures
https://resolver.caltech.edu/CaltechETD:etd-04202005-172426
Year: 2005
DOI: 10.7907/XMW0-DJ25
<p>Dense arrays of aligned carbon nanotubes are designed into strips — nanowicks — as a miniature wicking element for liquid delivery and potential microfluidic chemical analysis devices. The delivery function of nanowicks enables novel fluid transport devices to run without any power input, moving parts or external pump. The intrinsically nanofibrous structure of nanowicks provides a sieving matrix for molecular separations, and a high surface-to-volume ratio porous bed to carry catalysts or reactive agents.</p>
<p>This work also experimentally studies the spontaneous fluid transport along nanowicks. Liquid is conveyed through corner flow, surface flow, and interstitial flow through capillary force and the Marangoni effect. The main course for corner flow and surface flow follows Washburn behavior, and can deliver liquid centimeters away from the input blob with a speed on the order of millimeters per second depending on the nanowick configuration and the amount of input liquid. Corner flow can be minimized and even eliminated through proper nanowick and input design. Otherwise, corner flow interacts with surface flow in the first 2mm of the pathway closest to the input point. Interstitial flow dominates the late stage. It is driven by both capillary force and concentration-gradient-induced Marangoni force. The concentration gradient is determined by two competing rates: surfactant diffusion in solution and adsorption onto nanotube surfaces. The flow inside nanowicks may wick hundreds of microns in seconds or tens of seconds. A non-conventional advancing front may develop in the flow around nanowicks. They are seen as (i) Rayleigh instability-induced fingering in surface flow on millimeter-wide nanowicks, (ii) viscous instability-induced branching near almost-stagnant surface film at low surfactant concentration, and (iii) disjointed wetting domains at very low concentration.</p>https://resolver.caltech.edu/CaltechETD:etd-04202005-172426Bio-Inspired Visuomotor Convergence in Navigation and Flight Control Systems
https://resolver.caltech.edu/CaltechETD:etd-06072005-163739
Year: 2005
DOI: 10.7907/T5QZ-QS18
Insects exhibit incredibly robust closed loop flight dynamics in the face of uncertainties. A fundamental principle contributing to this unparalleled behavior is rapid processing and convergence of visual sensory information to flight motor commands via spatial wide-field integration, accomplished by retinal motion pattern sensitive interneurons (LPTCs) in the lobula plate portion of the visual ganglia. Within a control-theoretic framework, an inner product model for wide-field integration of retinal image flow is developed, representing the spatial decompositions performed by LPTCs in the insect visuomotor system. A rigorous characterization of the information available from this visuomotor convergence technique for motion within environments exhibiting non-homogeneous spatial distributions is performed, establishing the connection between retinal motion sensitivity shape and closed loop behavior. The proposed output feedback methodology is shown to be sufficient to give rise to experimentally observed insect navigational heuristics, including forward speed regulation, obstacle avoidance, hovering, and terrain following behaviors. Hence, extraction of global retinal motion cues through computationally efficient wide-field integration processing provides a novel and promising methodology for utilizing visual sensory information in autonomous robotic navigation and flight control applications.
https://resolver.caltech.edu/CaltechETD:etd-06072005-163739Unsteady Fluid Mechanics of Starting-Flow Vortex Generators with Time-Dependent Boundary Conditions
https://resolver.caltech.edu/CaltechETD:etd-04112005-151435
Year: 2005
DOI: 10.7907/QV8Y-YZ12
<p>Nature has repeatedly converged on the use of starting flows for mass, momentum, and energy transport. The vortex loops that form during flow initiation have been reproduced in the laboratory and have been shown to make a proportionally larger contribution to fluid transport than an equivalent steady jet. However, physical processes limit growth of the vortex loops, suggesting that these flows may be amenable to optimization. Although it has been speculated that optimal vortex formation might occur naturally in biological systems, previous efforts to quantify the biological mechanisms of vortex formation have been inconclusive. In addition, the unsteady fluid dynamical effects associated with starting flow vortex generators are poorly understood.</p>
<p>This thesis describes a combination of new experimental techniques and in vivo animal measurements that determine the effects of fluid-structure interactions on vortex formation by starting flow propulsors. Results indicate that vortex formation across various biological systems is manipulated by these kinematics in order to maximize thrust and/or propulsive efficiency. An emphasis on observed vortex dynamics and transient boundary conditions facilitates quantitative comparisons across fluid transport schemes, irrespective of their individual biological functions and physical scales.</p>
<p>The primary contributions of this thesis are the achievement of quantitative measures of unsteady vortex dynamics via fluid entrainment and added-mass effects, and the development of a robust framework to facilitate the discovery of general design principles for effective fluid transport in engineering technologies and biological therapies. The utility of this new research paradigm is demonstrated through a variety of examples.</p>https://resolver.caltech.edu/CaltechETD:etd-04112005-151435An Experimental Analysis of the Characteristic Behaviors of an Impedance Pump
https://resolver.caltech.edu/CaltechETD:etd-05232005-141405
Year: 2005
DOI: 10.7907/SQQK-FD11
<p>When a fluid-filled pliant tube is connected to tubing of a different impedance, a net flow in either direction can be induced by periodically compressing the pliant section asymmetrically from the ends. An experimental analysis of the characteristic behaviors of such a pump has been done demonstrating interesting results not predicted by prior analytical and computational results. Measurements show a complex non-linear behavior in response to the compression frequency, including distinct resonance peaks and reversals in flow direction. Ultrasound imaging provided a unique view of the tube wall and flow within, allowing us to visualize the wave propagation and reflection. Measurements include transient responses, resonant responses, and bulk flow behaviors for a variety of configurations. Net flow rates can exceed the volumetric displacement done by active compression demonstrating that, as a first approximation, this pump can have a higher efficiency than peristaltic pumping. Elasticity has been shown not to be a necessary factor in stimulating net forward flow.</p>
<p>Results from this study have helped show that a zebrafish (a model for human cardiac development) may utilize impedance pumping to drive circulation in early embryonic stages prior to valve formation as opposed to peristaltic pumping as was once thought. Additional research is being conducted to develop a micro-scaled version with applications in medicine, heat transfer, lab-on-chip technology, and micro-mixing.</p>https://resolver.caltech.edu/CaltechETD:etd-05232005-141405Cognitive Neural Prosthetics: Brain Machine Interfaces Based in Parietal Cortex
https://resolver.caltech.edu/CaltechETD:etd-06032005-170438
Year: 2005
DOI: 10.7907/V2BZ-EG30
Systems neuroscience has recently emerged as an applied field in the form of neural prosthetic development. This integration of empirical systems neuroscience with engineering in order to develop functional interfaces between external devices and the brain has not only been beneficial in its applied goal, but has resulted in observations of scientific interest. The body of work presented here demonstrates the efficacy of two varieties of brain machine interfaces (BMIs) based in Parietal Cortex. The first using information about intended reaches present in action potentials, the second using local field potentials (LFPs). Both studies were predicated and succeeded with offline analyses demonstrating feasibility and novel insight to the function and neural coding properties of Parietal Cortex. We found that using BMIs resulted in adaptive change which tended to improve performance. LFPs, though less successful than spikes for BMI control under these experimental conditions, appear to have a multiplexing of different types of information that might aid in BMIs as well as providing a different way of looking at the neural processing. A preliminary exploration of relative timing of spikes and LFPs might result in some of the adaptive properties observed during BMI use via spike timing dependent plasticity concludes the research presented here.https://resolver.caltech.edu/CaltechETD:etd-06032005-170438Integrated Parylene LC-ESI on a Chip
https://resolver.caltech.edu/CaltechETD:etd-03032005-135900
Year: 2005
DOI: 10.7907/00KQ-V723
<p>In this thesis, several microfluidic devices will be introduced to demonstrate the integration capability of a multilayer parylene surface micromachining technology. Due to its flexibility and versatility, various devices have been developed and integrated onto a single ship. Based on the technology, on-chip LC-ESI was successfully demonstrated.</p>
<p>Based on the technology, an electrostatically actuated micro peristaltic pump has been developed. An AC actuation voltage combined with a peristaltic actuation was used to demonstrate fluid pumping. A reasonable flow rate and pumping pressure were achieved. The pump dynamics and performance were then addressed further by an analysis based on a lumped-parameter model of the system.</p>
<p>Based on the same technology, an entirely surface micromachined electrostatically actuated valve has been demonstrated. A thermal flow sensor was integrated with the valve to be used for feedback control. Two modes, actuation voltage adjustment and PWM were investigated in characterizing the valve to control air flow. The testing results show that PWM has better linearity and performance.</p>
<p>Three types of capacitive fluidic sensors were demonstrated in several microfluidic applications. These include sensors for fluid pressure, flow rate, volume, and composition measurement. The sensors showed great promise for microfluidic applications because of their high sensitivity and easy integration capabilities. The integration of these sensors with abovementioned devices was achieved.</p>
<p>A novel electrochemical pumping system for on-chip LC gradient generation was demonstrated. This pump was able to deliver significant flow rates under high back pressures that are sufficient for many LC applications. On-chip gradient formation with integrated electrospray ionization was demonstrated.</p>
<p>Finally, a complete LC-ESI system was integrated in a chip format. Typical nano-LC reversed-phase gradient elution was demonstrated using on-chip electrolysis pump. Separated analytes from on-chip column were then sprayed into MS for analysis through an integrated ESI-nozzle. Separation results are comparable to those of commercial system. Peptide identification performance using the LC-ESI chip with MS was also very close to those achieved by the commercial system.</p>https://resolver.caltech.edu/CaltechETD:etd-03032005-135900Neural Computation of Self-Motion from Optic Flow in Primate Visual Cortex
https://resolver.caltech.edu/CaltechETD:etd-05242006-221959
Year: 2006
DOI: 10.7907/ahre-qy74
<p>Area MSTd is involved in the computation of heading direction from the focus of expansion (FOE) of the visual image. Our laboratory previously found that MSTd neurons adjust their focus tuning curves to compensate for shifts in the FOE produced by eye rotation (Bradley et al., 1996) as well as for changes in pursuit speed (Shenoy et al., 2002). The translation speed of an observer also affects the shift of the FOE. To investigate whether MSTd neurons can adjust their focus tuning curves to compensate for varying translation speeds, we recorded extracellular responses from 93 focus-tuned MSTd neurons in two rhesus monkeys (Macaca mulatta) performing pursuit eye movements across displays of varying translation speeds. We found that MSTd neurons had larger shifts in their tuning curves for slow translation speeds and smaller shifts for fast translation speeds. These shifts aligned the focus tuning curves with the true heading direction and not with the retinal position of the FOE. These results indicate that retinal cues related both to translation speed and extraretinal signals from pursuit eye movements are used by MSTd neurons to compute heading direction.</p>
<p>Although there is much evidence that MSTd neurons are involved in heading computation, it was not known in which coordinate frame the tuning curves were represented. We performed a second set of experiments to determine whether focus tuning curves in MSTd were represented in eye, head, body, or world coordinates. The coordinate frame was determined while the eyes were stationary (fixed gaze, simulated pursuit condition) and while the eyes were moving (real pursuit condition). We recorded extracellular responses from 80 MSTd neurons and found that the FOE tuning curves of the overwhelming majority of neurons were aligned in an eye-centered coordinate frame as opposed to head, body, or world-centered coordinates (fixed gaze: 77/80 (96%); real pursuit: 77/80 (96%); simulated pursuit 74/80 (93%); t-test, p<0.05). We also found that area MSTd demonstrated significant eye position gain modulation much like its posterior parietal neighbors. This gain modulation may be a method of transforming eye coordinates into other coordinate frames at later stations of the nervous system.</p>https://resolver.caltech.edu/CaltechETD:etd-05242006-221959System Architectures and Environment Modeling for High-Speed Autonomous Navigation
https://resolver.caltech.edu/CaltechETD:etd-05242006-190748
Year: 2006
DOI: 10.7907/8HT2-N165
<p>Successful high-speed autonomous navigation requires integration of tools from robotics, control theory, computer vision, and systems engineering. This thesis presents work that develops and combines these tools in the context of navigating desert terrain.</p>
<p>A comparative analysis of reactive, behavior-based, and deliberative control architectures provides important guidelines for design of robotic systems. These guidelines depend on the particular task and environment of the vehicle. Two important factors are identified which guide an effective choice between these architectures: dynamic feasibility for the vehicle, and predictability of the environment. This is demonstrated by parallels to control theory, illustrative examples, simulations, and analysis of Bob and Alice---Caltech's full-scale autonomous ground vehicle entries in the 2004 and 2005 Grand Challenge races, respectively.</p>
<p>Further, new model-based methods are developed for constructing and maintaining estimates of terrain elevation and road geometry. These are demonstrated in simulation and in fully autonomous operation of Alice, including accurate detection and tracking of the centerline of desert roads at speeds up to 5 m/s. Finally, Alice's navigation architecture is presented in full along with experimental results that demonstrate its capabilities.</p>https://resolver.caltech.edu/CaltechETD:etd-05242006-190748Multi-robot Systems: Modeling Swarm Dynamics and Designing Inspection Planning Algorithms
https://resolver.caltech.edu/CaltechETD:etd-05192006-063455
Year: 2006
DOI: 10.7907/G1T2-FB74
<p>For a variety of applications, the capability of simultaneous sensing and action in multiple locations that is inherent to multi-robot approaches offers potential advantages over single robot systems in robustness, efficiency, and application feasibility.</p>
<p>At the fully distributed and reactive end of the multi-robot system spectrum, I present mathematical modeling methodologies developed to predict and optimize a self-organized robotic swarm’s performance for several tasks. These models allow us to better understand the relationship between agent and group behavior by capturing the dynamics of these highly stochastic, nonlinear, asynchronous systems at various levels of abstraction, in some cases even achieving mathematical tractability. The models deliver qualitatively and quantitatively correct predictions several orders of magnitude more quickly than an embodied simulator can. Swarm modeling lays the foundation for more generalized SI system design methodology by saving time, enabling generalization to different robotic platforms, and estimating optimal design and control parameters.</p>
<p>In considering more complex target tasks and behaviors, efficiency and completeness of execution may be of concern, and a swarm approach may not be appropriate. In such cases a more deliberative approach may be warranted. In that context, I introduce the multi-robot boundary coverage problem, in which a group of robots is required to completely inspect the boundary of all two-dimensional objects in a specified environment. To make such a guarantee, I present a centralized planning approach that constructs a two-component abstraction of the problem: a graph representing the particular instance of the inspection task and a graph problem whose solution represents a complete plan for inspection. Using the building blocks of this approach, related inspection tasks that require the robotic system to adapt to a changes in team size and task assignment are also explored. The application of these planning methods to the case of long-term deployment for surveillance applications that require repetitive coverage is also discussed.</p>
<p>The recurring theme of this thesis is that we must look beyond implementation and validation of a particular system and ask how its design can contribute to the development of a more general design methodology.</p>https://resolver.caltech.edu/CaltechETD:etd-05192006-063455A Control System for Positioning Recording Electrodes to Isolate Neurons in Extracellular Recordings
https://resolver.caltech.edu/CaltechETD:etd-06042006-160620
Year: 2006
DOI: 10.7907/6HHC-5456
<p>This thesis presents an algorithm that autonomously positions recording electrodes inside cortical tissue so as to isolate and then maintain optimal extracellular signal recording quality without human intervention. The algorithm is used to improve the quality and efficiency of acute (daily insertion) recordings that are needed for basic research in neurophysiology. It also offers the potential to increase the longevity and quality of chronic (long-term implant) recordings by controlling an emerging class of chronic arrays in which the electrodes can be continually repositioned after implantation.</p>
<p>The challenges encountered in attempting to isolate neurons are studied. A solution is proposed in which a finite state machine oversees a number of signal processing steps, computes various metrics of the recording quality and issues commands to move the electrode close to neurons without causing them damage. A number of metrics of the quality of neuron isolation are compared.</p>
<p>The algorithm has been used to control a number of commercial microdrive systems, including a single-electrode FHC microdrive and multielectrode microdrives from Thomas Recording and NAN, as well as a novel miniature microdrive. The autonomous positioning software is used by several neuroscientists to perform basic neurophysiology research. Analysis of the system's performance in isolating neurons is included.</p>https://resolver.caltech.edu/CaltechETD:etd-06042006-160620Engineering Design Synthesis of Sensor and Control Systems for Intelligent Vehicles
https://resolver.caltech.edu/CaltechETD:etd-05252006-221412
Year: 2006
DOI: 10.7907/NB6H-S822
<p>This thesis investigates the application of formal engineering design synthesis methodologies to the development of sensor and control systems for intelligent vehicles.</p>
<p>A formal engineering design synthesis methodology based on evolutionary computation is presented, with special emphasis on dealing with modern engineering design challenges, such as high or variable complexity of design solutions, multiple conflicting design objectives, and noisy evaluation results, etc. The efficacy of the evolutionary design synthesis method is validated through multiple different case studies, where a variety of novel design solutions are generated to represent different engineering design trade-offs, and they have achieved performances comparable to, if not better than, that of hand-coded solutions in the same simplified environment. More importantly, this automatic design synthesis method shows great potential to handle more complex design problems, where a good hand-coded solution may be very difficult or even impossible to obtain. Moreover, the evolutionary design synthesis methodology appears promising to deal with uncertainty in the problem efficiently and adapt to the collective task nature well.</p>
<p>In addition, multiple levels of vehicle simulation models with different computational cost and fidelity as well as necessary driver behaviors are implemented for different types of simulation experiments conducted for different research purposes. Efforts are made to try to generate good candidate solutions efficiently with less computational time and human engineering effort.</p>
<p>Furthermore, a new threat assessment measure, time-to-last-second-braking (<i>T<sub>lsb</sub></i>), is proposed, which directly characterizes human natural judgment of the urgency and severity of threats in terms of time. Based on driver reaction time experimental results, new warning and overriding criteria are proposed in terms of the new <i>T<sub>lsb</sub></i> measure, and the performance is analyzed statistically in terms of two typical sample pre-crash traffic scenarios. Less affected by driver behavior variability, the new criteria characterize the current dynamic situations better than the previous ones, providing more appropriate warning and more effective overriding at the last moment. Finally, the possibility of frontal collision avoidance through steering (lane-changing) is discussed, and similarly the time-to-last-second-steering (<i>T<sub>lss</sub></i>) measure is proposed and compared with <i>T<sub>lsb</sub></i>.</p>https://resolver.caltech.edu/CaltechETD:etd-05252006-221412Algorithms for Mobile Robot Localization and Mapping, Incorporating Detailed Noise Modeling and Multi-scale Feature Extraction
https://resolver.caltech.edu/CaltechETD:etd-05262006-130209
Year: 2006
DOI: 10.7907/FN3J-M568
<p>Mobile robot localization and mapping in unknown environments is a fundamental requirement for effective autonomous navigation. Three different approaches to localization and mapping are presented. Each is based on data collected from a robot using a dense range scanner to generate a planar representation of the surrounding environment. This externally sensed range data is then overlayed and correlated to estimate the robot's position and build a map.</p>
<p>The three approaches differ in the choice of representation of the range data, but all achieve improvements over prior work using detailed sensor modeling and rigorous bookkeeping of the modeled uncertainty in the estimation processes. In the first approach, the raw range data points collected from two different positions are individually weighted and aligned to estimate the relative robot displacement. In the second approach, line segment features are extracted from the raw point data and are used as the basis for efficient and robust global map construction and localization. In the third approach, a new multi-scale data representation is introduced. New methods of localization and mapping are developed, taking advantage of this multi-scale representation to achieve significant improvements in computational complexity. A central focus of all three approaches is the determination of accurate and robust solutions to the data association problem, which is critical to the accuracy of any sensor-based localization and mapping method.</p>
<p>Experiments using data collected from a Sick LMS-200 laser scanner illustrate the effectiveness of the algorithms and improvements over prior work. All methods are capable of being run in real time on a mobile robot, and can be used to support fully autonomous navigation applications.</p>https://resolver.caltech.edu/CaltechETD:etd-05262006-130209A Compact System for Self-Motion Estimation
https://resolver.caltech.edu/CaltechETD:etd-05252006-145119
Year: 2006
DOI: 10.7907/50B3-RW79
Self-motion estimation is a vital problem for autonomous robots, frequently and appropriately addressed by vision algorithms. Most approaches involve repeating some local calculation over the entire imaging array, such as detection of locally salient features. A simple and local calculation can be efficiently implemented on the same chip as the photo-sensing array, thus parallelizing a huge computational task and vastly reducing the amount of data to transmit off chip. Mismatch between devices has previously been a stumbling block to producing truly useful arrays of local processing elements. Floating gate technology is used here as a compact means of programming away offsets in subcircuits to remedy this problem. A custom analog chip performs the above functions. For each pixel, the chip outputs sensed light intensity, the values of the vertical and horizontal intensity gradients, and a binary value indicating whether a feature is centered on that pixel. These values can be used as inputs to a motion estimation algorithm implemented on a standard computer.https://resolver.caltech.edu/CaltechETD:etd-05252006-145119Distributed Gradient Systems and Dynamic Coordination
https://resolver.caltech.edu/CaltechETD:etd-06262006-171822
Year: 2006
DOI: 10.7907/D1NJ-KF96
<p>Many systems comprised of interconnected sub-units exhibit coordinated behaviors; social groups, networked computers, financial markets, and numerous biological systems come to mind. There has been long-standing interest in developing a scientific understanding of coordination, both for explanatory power in the natural and economic sciences, and also for constructive power in engineering and applied sciences. This thesis is an abstract study of coordination, focused on developing a systematic "design theory" for producing interconnected systems with specifiable coordinated behavior; this is in contrast to the bulk of previous work on this subject, in which any design component has been primarily ad-hoc.</p>
<p>The main theoretical contribution of this work is a geometric formalism in which to cast distributed systems. This has numerous advantages and "naturally" parametrizes a wide class of distributed interaction mechanisms in a uniform way. We make use of this framework to present a model for distributed optimization, and we introduce the distributed gradient as a general design tool for synthesizing dynamics for distributed systems. The distributed optimization model is a useful abstraction in its own right and motivates a definition for a distributed extremum. As one might expect, the distributed gradient is zero at a distributed extremum, and the dynamics of a distributed gradient flow must converge to a distributed extremum. This forms the basis for a wide variety of designs, and we are in fact able to recover a widely studied distributed averaging algorithm as a very special case.</p>
<p>We also make use of our geometric model to introduce the notion of coordination capacity; intuitively, this is an upper bound on the "complexity" of coordination that is feasible given a particular distributed interaction structure. This gives intuitive results for local, distributed, and global control architectures, and allows formal statements to be made regarding the possibility of "solving" certain optimization problems under a particular distributed interaction model.</p>
<p>Finally, we present a number of applications to illustrate the theoretical approach presented; these range from "standard" distributed systems tasks (leader election and clock synchronization) to more exotic tasks like graph coloring, distributed account balancing, and distributed statistical computations.</p>https://resolver.caltech.edu/CaltechETD:etd-06262006-171822Temperature-Controlled Microchip Liquid Chromatography System
https://resolver.caltech.edu/CaltechETD:etd-04182006-162552
Year: 2006
DOI: 10.7907/ZDK5-Q871
<p>High-performance liquid chromatography (HPLC) is one of the most important analytical tools heavily used in the fields of chemistry, biotechnology, pharmaceutics, and the food industry. The power of liquid chromatography comes from its ability to achieve molecular separation with extremely high efficiency and its great flexibility of incorporating versatile sensors for detecting a broad range of analytes. In the past decades, great efforts have been put into liquid chromatography instrumentation and methods, aiming to further improve separation efficiency, sensitivity, repeatability, throughput, and costs. The contribution of this thesis is to illustrate with real examples the great potential of MEMS microchip liquid chromatography systems with on-chip temperature control for replacing and improving the conventional desktop HPLC systems.</p>
<p>This thesis is composed of seven chapters. Chapter 1 gives an introduction to MEMS technology and its application in making lab-on-a-chip systems. Chapter 2 describes the theoretical background and the evolution of HPLC technology. Chapter 3 demonstrates how to use state-of-the-art MEMS technology to make high-pressure microfluidic channels, which will be used for constructing microchip HPLC systems later. Chapter 4 describes a temperature-controlled microchip HPLC system that uses a temporal temperature gradient to achieve analyte elution. Separation of amino acids and low density lipoproteins was successfully demonstrated using the proposed system. Chapter 5 describes a novel embedded HPLC system, which demonstrated a record high pressure capacity (> 1000 psi) among microchip HPLC systems. High quality separation results of trace-level daunorubicin and doxorubicin were obtained using the proposed system and laser-induced fluorescence detection. A novel C4D sensor together with the RISE sensitivity enhancement method was proposed and investigated for the first time for microchip HPLC analyte detection. Chapter 6 describes the first work to pack 30 nm gold nanoparticles into the HPLC separation column as the stationary phase with the assistance of in-situ molecular self-assembly between nanoparticles and thiolated molecules. Preliminary results demonstrated the possibility of building fully filled nanoparticle HPLC columns for extremely high separation efficiency application. Chapter 7 then gives the conclusions of this thesis.</p>https://resolver.caltech.edu/CaltechETD:etd-04182006-162552Tools and Algorithms for Mobile Robot Navigation with Uncertain Localization
https://resolver.caltech.edu/CaltechETD:etd-06012006-150109
Year: 2006
DOI: 10.7907/R6YB-NQ21
The ability for a mobile robot to localize itself is a basic requirement for reliable long range autonomous navigation. This thesis introduces new tools and algorithms to aid in robot localization and navigation. I introduce a new range scan matching method that incorporates realistic sensor noise models. This method can be thought of as an improved form of odometry. Results show an order of magnitude of improvement over typical mobile robot odometry. In addition, I have created a new sensor-based planning algorithm where the robot follows the locally optimal path to the goal without exception, regardless of whether or not the path moves towards or temporarily away from the goal. The cost of a path is defined as the path length. This new algorithm, which I call "Optim-Bug," is complete and correct. Finally, I developed a new on-line motion planning procedure that determines a path to a goal that optimally allows the robot to localize itself at the goal. This algorithm is called "Uncertain Bug." In particular, the covariance of the robot's pose estimate at the goal is minimized. This characteristic increases the likelihood that the robot will actually be able to reach the desired goal, even when uncertainty corrupts its localization during movement along the path. The robot's path is chosen so that it can use known features in the environment to improve its localization. This thesis is a first step towards bringing the tools of mobile robot localization and mapping together with ideas from sensor-based motion planning.https://resolver.caltech.edu/CaltechETD:etd-06012006-150109A Compact System for Self-Motion Estimation
https://resolver.caltech.edu/CaltechETD:etd-05252006-145119
Year: 2006
DOI: 10.7907/50B3-RW79
Self-motion estimation is a vital problem for autonomous robots, frequently and appropriately addressed by vision algorithms. Most approaches involve repeating some local calculation over the entire imaging array, such as detection of locally salient features. A simple and local calculation can be efficiently implemented on the same chip as the photo-sensing array, thus parallelizing a huge computational task and vastly reducing the amount of data to transmit off chip. Mismatch between devices has previously been a stumbling block to producing truly useful arrays of local processing elements. Floating gate technology is used here as a compact means of programming away offsets in subcircuits to remedy this problem. A custom analog chip performs the above functions. For each pixel, the chip outputs sensed light intensity, the values of the vertical and horizontal intensity gradients, and a binary value indicating whether a feature is centered on that pixel. These values can be used as inputs to a motion estimation algorithm implemented on a standard computer.https://resolver.caltech.edu/CaltechETD:etd-05252006-145119Geometrical Analysis of Spatio-temporal Planning Problems
https://resolver.caltech.edu/CaltechETD:etd-05202007-135411
Year: 2007
DOI: 10.7907/917G-MJ20
In this thesis I represent and analyze spatially and temporally constrained multi-agent planning problems using tools from geometry and advanced calculus. The two problems considered in this thesis are multi-agent rendezvous and dynamic sensor coverage. Together, these problems encompass the cooperation, constraint representation,and task scheduling aspects of multi-agent planning problems. I have represented the constraint of the rendezvous problem on the phase space and shown that the fulfilment of rendezvous constraints is equivalent to certain conical regions being invariant. Alternatively, for the dynamic coverage problem, the constraints can be adequately represented on the uncertainty space and sensor motion laws can be obtained by partitioning the uncertainty space and making decisions based on which partition the uncertainty lies in. I have examined convergence behavior of sensor motion under such laws.https://resolver.caltech.edu/CaltechETD:etd-05202007-135411Information-Theoretic Methods for Modularity in Engineering Design
https://resolver.caltech.edu/CaltechETD:etd-05282007-183612
Year: 2007
DOI: 10.7907/ZBBK-JM82
<p>Due to their many advantages, modular structures commonly exist in artificial and natural systems, and the concept of modular product design has recently received extensive attention from the engineering research community. Although some work has been done on modularity, most of it is qualitative and exploratory in nature, and little is quantitative. One reason for this gap is the lack of a clear definition of modularity. This thesis begins with a detailed discussion on the concepts of “modularity” and “module.”</p>
<p>Based on the background presented here, a mutual information-based method is proposed to quantify modularity. The method is based on the view that coupling is information flow instead of real physical interactions. Information flow can be quantified by mutual information, which is based on randomness (or uncertainty). Since most engineering products can be modeled as stochastic systems and therefore have randomness, the mutual information-based method can be applied in very general cases, and it is shown that the commonly existing linkage counting modularity measure is a special case of the mutual information-based modularity measure.</p>
<p>The mutual information-based method is applicable to final design products. But at the early stage of the engineering design process, there are generally only function diagrams. To exploit the benefits of modularity as early as possible, a minimal description length principle-based modularity measure is proposed to determine the modularity of graph structures, which can represent function diagrams. The method is used as criteria to hierarchically decompose abstract graph structures and the real function structure of an HP printer by evolutionary computation. Due to the specialty of genome representations in evolutionary computation, new genetic operators are developed to determine optimal hierarchical decompositions.</p>
<p>This quantitative modularity measure has been developed to synthesize modular engineering products, especially by evolutionary design. There are many factors affecting evolving modular structures, such as genome representation, fitness function, learning, and task structure. The thesis preliminarily studies the effects of the modularity of tasks on the modularity of products in evolutionary computation. Using feed-forward neural networks as examples, the results show that the effects are task-dependent and rely on the amount of resources available for the tasks.</p>https://resolver.caltech.edu/CaltechETD:etd-05282007-183612Robotics Training Algorithms for Optimizing Motor Learning in Spinal Cord Injured Subjects
https://resolver.caltech.edu/CaltechETD:etd-08142006-165844
Year: 2007
DOI: 10.7907/EH12-WD80
<p>The circuitries within the spinal cord are remarkably robust and plastic. Even in the absence of supraspinal control, such circuitries are capable of generating functional movements and changing their level of excitability based on a specific combination of properceptive inputs going into the spinal cord. This has led to an increase in locomotor training, such as Body Weight Support Treadmill training (BWST) for spinal cord injured (SCI) patients. However, today, little is known about the underlying physiological mechanisms responsible for the locomotor recovery achieved with this type of rehabilitative training, and the optimal rehabilitative strategy is still unknown.</p>
<p>This thesis describes a mouse model to study the effect of rehabilitative training on SCI. Using this model, the effects of locomotor recovery on adult spinal mice following complete spinal cord transaction is examined. Results that indicate adult spinal mice can be robotically trained to step, and when combined with the administration of quipazine (a broad serotonin agonist), there is an interaction and retention effect. Results also demonstrate that the training paradigm can be optimized in using “Assisted-as-Needed” (AAN) training. To find the optimal AAN training parameters, a learning model is developed to test the effect of various parameters of the AAN training algorithm. Simulation results from our model show that learning is training-dependent. In addition, the model predicts that improved motor learning can improve post-SCI by making the AAN training more adaptable.</p>
<p>The primary contributions of this thesis are twofold, in biology and engineering. We develop a mouse model using novel robotic devices and controls that can be used to study SCI and other locomotor disorders in the future by taking advantage of the many different strains of transgenic mice that are commercially available. We also further confirm that sensory integration responsible for motor control is distributed throughout the hierarchy of the neuromuscular system and can be achieved within the isolated spinal cord. Lastly, by developing a learning model, we can start looking into how variability plays a role in motor learning, the understanding of which will have profound implications in neurophysiology, machine learning and adaptive optimal controls research.</p>https://resolver.caltech.edu/CaltechETD:etd-08142006-165844Exploration into the Feasibility of Underwater Synthetic Jet Propulsion
https://resolver.caltech.edu/CaltechETD:etd-09252006-134742
Year: 2007
DOI: 10.7907/72SZ-T823
<p>This thesis explores the feasibility of using synthetic jet actuators for the propulsion of small underwater vehicles. This work was inspired by the widespread use of pusatile jet propulsion by sea creatures such as squid, salp, and jellyfish. The jets created by these animals utilize vortex rings for thrust production. A method for creating similar vortex ring-based jets is the use of synthetic, or zero net mass flux, jets. These jets, which form a jet structure through the alternating sucking and blowing of fluid through a single orifice, have previously been investigated for the utility in air flow control.</p>
<p>The design, construction, and testing of aquatic synthetic jet prototypes is presented. Force measurement and flow visualization experiments are performed on these jets to gain an understanding of the forces and flow structures produced. The flow visualizations confirm the outflow vortex ring observations reported previously in the literature and present the first images of vortex ring formation inside the synthetic jet chamber. A new phenomenon, that of self-induced coflow upstream of the jet orifice, is discussed. The force measurements present confirmation that a net thrust is produced by the jets and give insight to the relationship between jet forcing parameters (such as frequency) and the resulting thrust. An automated genetic algorithmic approach to optimizing the thrust for a given jet geometry is also presented and tested.</p>
<p>Using the results of these experiments I propose a model for synthetic jet thrust. This model asserts that there are three force producing components to the flow: orifice inflow, orifice outflow, and a self-induced coflow. The contribution of each of these components is derived and compared with experimental results.</p>
<p>Included at the end of this thesis is a preliminary study into possible vehicle architecture for the utilization of synthetic jet thrusters.</p>https://resolver.caltech.edu/CaltechETD:etd-09252006-134742Visually Mediated Control of Flight in Drosophila: Not Lost in Translation
https://resolver.caltech.edu/CaltechETD:etd-01082007-033253
Year: 2007
DOI: 10.7907/YYSN-7C82
<p>Flying insects exhibit stunning behavioral repertoires that are largely mediated by the visual control of flight. For this reason, presenting a controlled visual environment to tethered insects has been and continues to be a powerful tool for studying the sensory control of complex behaviors. The work presented in this dissertation concerns several robust behavioral responses exhibited by Drosophila that shed light on some of the challenges of visual navigation. To address questions of visual flight control in Drosophila, a modular display system has been designed and has proven to be a robust experimental instrument. The display system has enabled the wide variety of experimental paradigms presented in the thesis.</p>
<p>Much is known about the responses of tethered Drosophila to rotational stimuli. However, the processing of the more complex patterns of motion that occur during translatory flight is largely unknown. Recent experimental results have demonstrated that Drosophila turn away from visual patterns of expansion. However, the avoidance of expansion is so vigorous, that flies robustly orient towards the focus of contraction of a translating flow field. Much of the effort documented in this thesis has sought to explain this paradox.</p>
<p>The paradox has been largely resolved by several significant findings. When undergoing flight directed towards a prominent object, Drosophila will tolerate a level of expansion that would otherwise induce avoidance. The expansion-avoidance behavior is also critically dependent on the speed of image motion; in response to reduced speeds of expansion, Drosophila exhibit a centering response in which they steer towards the focus of expansion by balancing the image motion seen by both eyes. Taken together, these behaviors contribute to a model of Drosophila's visual flight control as emerging from multiple behavioral modules that operate concurrently.</p>
<p>Simple computational models of Drosophila's visual system are used to demonstrate that the experimental results arrived at by doing psychophysics on tethered animals actually yield sensible navigation strategies. This final component of the thesis documents an effort to close the feedback loop around the experimenter, by using computational models of Drosophila behavior to constrain the design of future experiments.</p>https://resolver.caltech.edu/CaltechETD:etd-01082007-033253Intelligent Information-Gathering: Using Control for Sensing and Decision-Making
https://resolver.caltech.edu/CaltechETD:etd-05312007-024822
Year: 2007
DOI: 10.7907/V5S4-4197
Information is everywhere and evolving, which necessitates both deliberate and efficient processing to acquire a good understanding of the dynamic situation, environment, or system of interest. Intelligent agents such as autonomous mobile sensors can control the way they gather information and thereby take advantage of feedback to improve the quality of that information. This approach reflects a shift from traditional "sensing for control" notions to "control for sensing" methods for addressing information-based objectives. This thesis presents several algorithms for distributed sensing tasks in the context of a team of mobile sensing agents. Applications of these types of mobile sensor networks include target tracking, dynamic environment monitoring, and distributed classification. These methods point beyond the use of sensory data for control and toward a framework for using control to improve information-based decisions made by intelligent agents. The sequential decision-theoretic framework presented herein has relevant applications in engineered systems such as search and rescue using a robotic team, as well as potential connections to natural systems including search strategies in the human vision system.https://resolver.caltech.edu/CaltechETD:etd-05312007-024822Application of Stochastic Simulation Methods to System Identification
https://resolver.caltech.edu/CaltechETD:etd-05222007-152843
Year: 2007
DOI: 10.7907/9JKZ-Z422
<p>Reliable predictive models for the response of structures are a necessity for many branches of earthquake engineering. However, the process of choosing an appropriate class of models to describe a system, known as model-class selection, and identifying the specific predictive model based on available data, known as system identification, is difficult. Variability in material properties, complex constitutive behavior, uncertainties in the excitations caused by earthquakes, and limited constraining information make system identification an ill-conditioned problem. In addition, model-class selection is not trivial, as it involves balancing predictive power with simplicity.</p>
<p>These problems of system identification and model-class selection may be addressed using a Bayesian probabilistic framework that provides a method for combining prior knowledge of a system with measured data and for choosing between competing model classes.</p>
<p>Similar approaches have been used in the field of system identification, but these methods (referred to as asymptotic-approximation-based methods) often focus on the model defined by the set of most plausible parameter values and have difficulty dealing with ill-conditioned problems, where there may be many models with high plausibility. It is demonstrated here that ill-conditioned problems in system identification and model-class selection can be effectively addressed using stochastic simulation methods.</p>
<p>This work focuses on the application of stochastic simulation to updating and comparing model classes in problems of: (1) development of empirical ground motion attenuation relations, (2) structural model updating using modal data for the purposes of structural health monitoring, and (3) identification of hysteretic structural models, including degrading models, from seismic structural response.</p>
<p>In cases where asymptotic approximation-based methods are appropriate, the results obtained using stochastic simulation show good agreement with results from asymptotics. For cases involving ill-conditioned problems based on simulated data, stochastic simulation methods are successfully applied to obtain results in situations where the use of asymptotics is infeasible. Finally, preliminary studies using stochastic simulation to identify a deteriorating hysteretic model with sparse real data from a structure damaged in an earthquake show that the high-plausibility models demonstrate behavior consistent with the observed damage, indicating that there is promise in applying these methods to ill-conditioned problems in the real world.</p>https://resolver.caltech.edu/CaltechETD:etd-05222007-152843Coordinated Control for Networked Multi-Agent Systems
https://resolver.caltech.edu/CaltechETD:etd-09182006-162259
Year: 2007
DOI: 10.7907/ZRAD-XN95
<p>Coordination in networked multi-agent systems attracts significant interest in the realm of engineering. Typical examples include formations of unmanned aerial vehicles, automated highway systems, and sensor networks. One common feature for these systems is that coordinated behaviors are exhibited by interactions among agents where information exchange and manipulation are necessary. In this work, three relevant issues are investigated in detail: uniform strategy for multi-agent formation control, fast-converging consensus protocols, and packet-based state estimation over communication networks.</p>
<p>Formation control of multi-agent systems involves harmony among local controller design, interaction topology analysis, and objective agreement among networked agents. We propose a novel control strategy so that each agent responds to neighbors' behaviors as well as acts towards the global goal. Thus, information flows for local interactions and global objective synchronization are studied separately. Using the tools from signal flow graphs and algebraic graph theory, we show that this new strategy eases the design of local controllers by relaxing stabilizing conditions. Robustness against the link failure and scalable disturbance resistance are also discussed based on small-gain theory. Experimental results on the Caltech multi-vehicle wireless testbed are provided to verify the feasibility and efficiency of this control strategy.</p>
<p>Consensus protocols over communication networks are used to achieve agreement among agents. One important issue is the convergence speed. We propose multi-hop relay protocols for fast consensus seeking. Without physically changing the topology of the communication network, this type of distributed protocol increases the algebraic connectivity by employing multi-hop paths in the network. We also investigate the convergence behaviors of consensus protocols with communication delays. It is interesting that, the faster the protocol converges, the more sensitive it is to the delay. This tradeoff is identified when we investigate delay margins of multi-hop relay protocols using the frequency sweep method.</p>
<p>Efficiently estimating the states of other agents over communication links is also discussed in this work. When information flows in the network, packet-based data is normally not retransmitted in order to satisfy real-time requirements. Thus, packet drops and random delays are inevitable. In this context, observation data that the estimator can receive is intermittent. In order to decrease the chance of losing packets and efficiently using the limited bandwidth, we introduce multiple description source codes to manipulate the data before transmission. Using modified algebraic Riccati equations, we show that multiple description codes improve the performance of Kalman filters over a large set of packet-dropping scenarios. This problem is also generalized to the case where observation data has an independent and identical static distribution over a finite set of observation noise. Moreover, Kalman filtering with bursty packet drops is also discussed based on the two-state Markov chain model.</p>https://resolver.caltech.edu/CaltechETD:etd-09182006-162259Automated Design Synthesis of Structures using Growth Enhanced Evolution
https://resolver.caltech.edu/CaltechETD:etd-06102008-153834
Year: 2008
DOI: 10.7907/28H7-E831
<p>Engineering design is a complex problem on generating and evaluating a variety of options. In traditional methods, this typically involves evaluating up to a dozen different point designs. The limit on the process is the amount of time to generate, refine, and evaluate the various concepts. Using a computer helps to speed up the process, but human involvement still remains the weakest link.</p>
<p>The natural extension of this process is to continually and rapid generate, refine, and evaluate concepts entirely automatically. Evolutionary Algorithms provide such a method, by emulating natural evolution. The computer maintains a population point design, each of which is represented by a gene string that is allowed to change (mutate) and combine with other genes (crossover). At each generation, every individual is modified then evaluated and the improved solutions proceed to the next generation.</p>
<p>This thesis will extend the biological model by introducing a growth process to each individual. This is akin to the concept of a multi-cellular organism developing in the womb. An encoding for discrete truss structures is described that provides for such an extension. The truss grows from a few basic elements. After showing several examples demonstrating the growth process, the method is applied to a couple simple examples using evolutionary algorithms.</p>https://resolver.caltech.edu/CaltechETD:etd-06102008-153834Managing Information in Networked and Multi-Agent Control Systems
https://resolver.caltech.edu/CaltechETD:etd-12192007-153619
Year: 2008
DOI: 10.7907/84NT-9N46
Traditional feedback control systems give little attention to issues associated with the flow of information through the feedback loop. Typically implemented with dedicated communication links that deliver nearly precise, reliable and non-delayed information, researchers have not needed to concern themselves with issues related to quantized, delayed and even lost information. With the advent of newer technologies and application areas that pass information through non-reliable networks, these issues can not be ignored. In recent years the field of Networked Control Systems (NCS) has emerged to describe situations where these issues are present. The research in this field focuses on quantifying performance degradations in the presence of network effects and proposing algorithms for managing the information flow to counter those negative effects. In this thesis I propose and analyze algorithms for managing information flow for several NCS scenarios; state estimation with lossy measurement signals, using input buffers to reduce the frequency of communication with a remote plant, and performing state estimation when control signals are transmitted to a remote plant via a lossy communication link with no acknowledgement signal at the estimator. Multi-agent coordinated control systems serve as a prime example of an emerging area of feedback control systems that utilize feedback loops with information passed through possibly imperfect communication networks. In these systems, agents use a communication network to exchange information in order to achieve a desired global objective. Hence managing the information flow has a direct impact on the performance of the system. I also explore this area by focusing on the problem of multi-agent average consensus. I propose an algorithm based on a hierarchical decomposition of the communication topology to speed up the time to convergence. For all these topics I focus on designing intuitive algorithms that intelligently manage the information flow and provide analysis and simulations to illustrate their effectiveness.
https://resolver.caltech.edu/CaltechETD:etd-12192007-153619Stochastic System Design and Applications to Stochastically Robust Structural Control
https://resolver.caltech.edu/CaltechETD:etd-12062007-135604
Year: 2008
DOI: 10.7907/6TRA-1C83
<p>The knowledge about a planned system in engineering design applications is never complete. Often, a probabilistic quantification of the uncertainty arising from this missing information is warranted in order to efficiently incorporate our partial knowledge about the system and its environment into their respective models. In this framework, the design objective is typically related to the expected value of a system performance measure, such as reliability or expected life-cycle cost. This system design process is called stochastic system design and the associated design optimization problem stochastic optimization. In this thesis general stochastic system design problems are discussed. Application of this design approach to the specific field of structural control is considered for developing a robust-to-uncertainties nonlinear controller synthesis methodology.</p>
<p>Initially problems that involve relatively simple models are discussed. Analytical approximations, motivated by the simplicity of the models adopted, are discussed for evaluating the system performance and efficiently performing the stochastic optimization. Special focus is given in this setting on the design of control laws for linear structural systems with probabilistic model uncertainty, under stationary stochastic excitation. The analysis then shifts to complex systems, involving nonlinear models with high-dimensional uncertainties. To address this complexity in the model description stochastic simulation is suggested for evaluating the performance objectives. This simulation-based approach addresses adequately all important characteristics of the system but makes the associated design optimization challenging. A novel algorithm, called Stochastic Subset Optimization (SSO), is developed for efficiently exploring the sensitivity of the objective function to the design variables and iteratively identifying a subset of the original design space that has high plausibility of containing the optimal design variables. An efficient two-stage framework for the stochastic optimization is then discussed combining SSO with some other stochastic search algorithm. Topics related to the combination of the two different stages for overall enhanced efficiency of the optimization process are discussed.</p>
<p>Applications to general structural design problems as well as structural control problems are finally considered. The design objectives in these problems are the reliability of the system and the life-cycle cost. For the latter case, instead of approximating the damages from future earthquakes in terms of the reliability of the structure, as typically performed in past research efforts, an accurate methodology is presented for estimating this cost; this methodology uses the nonlinear response of the structure under a given excitation to estimate the damages in a detailed, component level.</p>
https://resolver.caltech.edu/CaltechETD:etd-12062007-135604Parylene Technology for Neural Probes Applications
https://resolver.caltech.edu/CaltechETD:etd-11262007-125539
Year: 2008
DOI: 10.7907/GH99-K875
<p>Neural probes are important tools in detecting and studying neuron activities. Although people have been working on neural probe development for a long time, the current neural probes (including metal-wire probes and silicon neural probes) are still far from being satisfactory. An ideal neural probe array should have good biocompatibility, high-density electrodes with high signal-to-noise ratio, flexible cables for interconnections, integrated electronics, and even integrated actuators to track neuron movement.</p>
<p>The work of this thesis focused on applying parylene technology to neural probes development to make a new generation of neural probes with better functions. With the properties of high electrical resistivity, mechanical flexibility, biocompatibility, low coefficient of friction, and an easy deposition/etching process, parylene is a good material for neural probe applications. In this thesis, we have designed, fabricated, and characterized a new parylene neural probe with a long, flexible parylene cable for a neural prosthesis system. Parylene layers are first used on the silicon probe shank with multiple electrodes as insulation and protective layers. And long parylene flexible cables are first monolithically integrated with silicon neural probes. A 96-electrode high-density, 3-D neural probe array for chronic implantation has been demonstrated. Different types of electrolysis actuators (including a silicon diaphragm actuator and a parylene balloon actuator) have been made and tested. The research on electrolysis-based actuators shows their great potential to be used for movable neural probes.</p>
<p>Compared with the traditional silicon neural probes (e.g., the Michigan probes, the Utah electrode arrays), our microfabricated neural probes have much longer and stronger probe shanks (8 or 12 mm long, able to penetrate the human pia) and much longer flexible parylene cable (about 7 or 12 cm, long enough to go through a percutaneous connector and the human skull). At the same time, our new probe arrays are shown to have better biocompatibility (being totally covered with parylene material), lower stress, better penetration ability, and greater flexibility for making high-density 3-D arrays and for use in chronic neural signal recording implantation.</p>
https://resolver.caltech.edu/CaltechETD:etd-11262007-125539Development of Flexible Parylene-based Microtechnologies for Retinal and Spinal Cord Stimulation and Recording
https://resolver.caltech.edu/CaltechETD:etd-11132007-223407
Year: 2008
DOI: 10.7907/ZCXK-8T38
<p>The problems of outer retinal degeneration and spinal cord injury affect millions of people worldwide, often resulting in devastating blindness and para- or quadriplegia that strongly impair a person’s activities of daily living and impact their level of happiness. To help thwart the effects of these diseases, novel flexible parylene-based microtechnologies have been developed for functional electrical stimulation and recording in retinal and spinal cord prosthetics. Microelectrode arrays have been microfabricated according to a single-metal-layer process and a revolutionary dual-metal-layer process that promises to meet the needs of extremely high-density stimulation applications. Arrays have been fabricated of thin-film platinum, electroplated platinum, and iridium, all on parylene substrates, some electrodes surviving for more than 430 million pulses without failing. In addition, a new annealing and heat-molding process has been implemented to improve parylene to parylene adhesion and conform electrode arrays to approximate the curvature of canine retinas. A chronic implantation study of the mechanical effects of parylene-based electrode arrays on the retina over a six month follow-up period has provided excellent results. Both retinal and spinal stimulation and recording from such arrays have been demonstrated.</p>
<p>The first packaging technology for high lead-count prostheses capable of fully scalable interconnection of a high-density electrode array, radiofrequency telemetry coils, and other discrete components such as chip capacitors, with prefabricated, stand-alone driver circuitry is also presented, combining the best features of chip-level and wafer-level packaging technologies. This parylene-based drop-chip technology enables application-specific integrated circuits (ASICs) to be directly integrated into the fabrication process of the other system components, such that the resulting device is flexible, facilitating surgical implantation. The ASIC-to-electrode interconnects are patterned using standard photolithography and standard microfabrication techniques, enabling the density of interconnects to scale to the limits of the lithographic equipment used to define the etch holes over the on-chip pads. Electrical test results verify the efficacy of this cost-effective packaging scheme, and pave the way for a monolithic implantable parylene-based prosthesis system, which has been designed. Surgical tests of monolithic geometries for all-intraocular retinal prostheses have been conducted, and an exciting new configuration for such a device has been discovered.</p>
https://resolver.caltech.edu/CaltechETD:etd-11132007-223407Automated Visual Tracking for Behavioral Analysis of Biological Model Organisms
https://resolver.caltech.edu/CaltechETD:etd-05272008-161801
Year: 2008
DOI: 10.7907/TSQ7-SN68
<p>Capturing the detailed motion and behavior of biological organisms plays an important role in a wide variety of research disciplines. Many studies in biomechanics, neuroethology, and developmental biology rely on analysis of video sequences to understand the underlying behavior. However, the efficient and rapid quantification of these complex behavioral traits imposes a major bottleneck on the elucidation of many interesting scientific questions. The goal of this thesis is to develop a suite of model-based visual tracking algorithms that will apply across a variety of model organisms used in biology. These automated tracking algorithms operate in a high-throughput, high-resolution manner needed for a productive synthesis with modern genetic approaches. To this end, I demonstrate automated estimation of the detailed body posture of nematodes, zebrafish, and fruit flies from calibrated video.</p>
<p>The current algorithm utilizes a generative geometric model to capture the organism's shape and appearance. To accurately predict the organism's motion between video frames, I incorporate a motion model that matches tracked motion patterns to patterns in a training set. This technique is invariant with respect to the organism's velocity and can easily incorporate training data from completely different motion patterns. The prediction of the motion model is refined using measurements from the image. In addition to high-contrast feature points, I introduce a region, segmentation model based on level sets that are formally integrated into the observation framework of an Iterated Kalman Filter (IKF). The prior knowledge provided by the geometric and motion models improves tracking accuracy in the presence of partial occlusions and misleading visual cues.</p>
<p>The method is used to track the position and shape of multiple nematodes during mating behavior, zebrafish of different ages during escape response, and fruit flies during take off maneuvers. These applications demonstrate the modular design of this model-based visual tracking system, where the user can specify which components are appropriate to a given experiment. In contrast to other approaches, which are customized to a particular organism or experimental setup, my approach provides a foundation that requires little re-engineering whenever the experimental parameters are changed.</p>https://resolver.caltech.edu/CaltechETD:etd-05272008-161801The Neurochip: A Complete System for Long-Term Investigation of Cultured Neural Network Connectivity
https://resolver.caltech.edu/CaltechETD:etd-04162008-113925
Year: 2008
DOI: 10.7907/EAMA-EQ86
<p>Traditional techniques for investigating cultured neural networks, such as the patch clamp and multi-electrode array, are limited by 1) the number of identified cells which can be in simultaneous electrical contact, 2) the length of time for which cells can be studied, and 3) the lack of 1:1 neuron-to-electrode specificity.</p>
<p>Here, I present a novel device --- dubbed the "neurochip" --- which overcomes these limitations. This micromachined device consists of 4x4 array of "neurocages" which mechanically trap a neuron near an extracellular electrode. While the cell body is trapped, the axon and dendrites can freely grow into the surrounding area to form a network. The electrode is bi-directional, capable of both stimulating and recording action potentials. This system is noninvasive, so that an entire network --- all constituent neurons --- can be studied over its lifetime with fixed 1:1 neuron-to-electrode correspondence. Proof-of-concept experiments have been completed to illustrate that functional networks do indeed form in the neurochip system, and that suprathreshold connectivity can be fully mapped over several weeks. The neurochip opens a new domain in neurobiology for studying small cultured neural networks.</p>
https://resolver.caltech.edu/CaltechETD:etd-04162008-113925Robotic Training for Motor Rehabilitation after Complete Spinal Cord Injury
https://resolver.caltech.edu/CaltechETD:etd-09202007-135027
Year: 2008
DOI: 10.7907/T01R-P904
<p>The spinal cord circuits have a great degree of automaticity and plasticity. They are able to generate complex locomotor patterns such as stepping and scratching even without input from supraspinal nervous systems. When provided with ensembles of afferent sensory information input associated with a specific motor task, e.g., stepping, the spinal cord can "learn" to perform that task even if it is isolated from the supraspinal nervous systems.</p>
<p>The plasticity of the spinal cord led researchers to study the use of physical locomotor training, e.g., treadmill step training with body weight support, to rehabilitate locomotor function after spinal cord injury (SCI). With intensive training, the spinal-cord-injured subject can recover some level of stepping ability. Explorations were made in this thesis to find an optimal training paradigm. Novel assist-as-needed paradigms were developed to allow variability during training since it is an intrinsic feature of normal stepping. Comparative experiments were conducted against fixed-trajectory training. Results demonstrated that variability is an important factor to induce more improvement in step training.</p>
<p>Standing is another important function in one's daily life, though it received less research attention than stepping. A prototype stand platform with 6 degrees of freedom was developed as an experimental tool for stand and postural study. Analogous to step training, we tested the effect of daily training on extensor responses in the hind limbs of complete spinal rats. The results showed no significant effect of the training. This led to the conclusion that without tonic input, the spinal cord has very limited ability to generate enough extensor muscle tone and to respond to postural disturbance. Further studies in standing rehabilitation should combine other methods to provide tonic inputs to the spinal cord.</p>https://resolver.caltech.edu/CaltechETD:etd-09202007-135027Robust Model Predictive Control with a Reactive Safety Mode
https://resolver.caltech.edu/CaltechETD:etd-05072008-131735
Year: 2008
DOI: 10.7907/S0VN-VE35
<p>Control algorithms suitable for online implementation in engineering applications, such as aerospace and mechanical vehicles, often require adherence to physical state and control constraints. Additionally, the chosen algorithms must provide robustness to uncertainty affecting both the system dynamics and the constraints. As further autonomy is built into these systems, the algorithms must be capable of blending multiple operational modes without violating the intrinsic constraints. Further, for real-time applications, the implemented control algorithms must be computationally efficient and reliable. The research in this thesis approaches these application needs by building upon the framework of MPC (Model Predictive Control).</p>
<p>The MPC algorithm makes use of a nominal dynamics model to predict and optimize the response of a system under the application of a feedforward control policy, which is computed online in a finite-horizon optimization problem. The MPC algorithm is quite general and can be applied to linear and nonlinear systems and include explicit state and control constraints. The finite-horizon optimization is advantageous given the finite online computational capabilities in practical applications. Further, recursively re-solving the finite-horizon optimization in a compressing- or receding-horizon manner provides a form of closed-loop control that updates the feedforward control policy by setting the nominal state at re-solve to the current actual state. However, uncertainty between the nominal model and the actual system dynamics, along with constraint uncertainty can cause feasibility, and hence, robustness issues with the traditional MPC algorithm.</p>
<p>In this thesis, an R-MPC (Robust and re-solvable MPC) algorithm is developed for uncertain nonlinear systems to address uncertainty affecting the dynamics. The R-MPC control policy consists of two components: the feedforward component that is solved online as in traditional MPC; and a separate feedback component that is determined offline, based on a characterization of the uncertainty between the nominal model and actual system. The addition of the feedback policy generates an invariant tube that ensures the actual system trajectories remain in the proximity of the nominal feedforward trajectory for all time. Further, this tube provides a means to theoretically guarantee continued feasibility and thus re-solvability of the R-MPC algorithm, both of which are required to guarantee asymptotic stability.</p>
<p>To address uncertainty affecting the state constraints, an SR-MPC (Safety-mode augmented R-MPC) algorithm is developed that blends a reactive safety mode with the R-MPC algorithm for uncertain nonlinear systems. The SR-MPC algorithm has two separate operational modes: standard mode implements a modified version of the R-MPC algorithm to ensure asymptotic convergence to the origin; safety mode, if activated, guarantees containment within an invariant set about a safety reference for all time. The standard mode modifies the R-MPC algorithm with a special constraint to ensure safety-mode availability at any time. The safety-mode control is provided by an offline designed control policy that can be activated at any time during standard mode. The separate, reactive safety mode provides robustness to unexpected state-constraint changes; e.g., other vehicles crossing/stopping in the feasible path, or unexpected ground proximity in landing scenarios.</p>
<p>Explicit design methods are provided for implementation of the R-MPC and SR-MPC algorithms on a class of systems with uncertain nonlinear terms that have norm-bounded derivatives. Further, a discrete SR-MPC algorithm is developed that is more broadly applicable to real engineering systems. The discrete algorithm is formulated as a second-order cone program that can be solved online in a computationally efficient manner by using interior-point algorithms, which provide convergence guarantees in finite time to a prescribed level of accuracy.</p>
<p>This discrete SR-MPC algorithm is demonstrated in simulation of a spacecraft descent toward a small asteroid where there is an uncertain gravity model, as well as errors in the expected surface altitude. Further, realistic effects such as control-input uncertainty, sensor noise, and unknown disturbances are included to further demonstrate the applicability of the discrete SR-MPC algorithm in a realistic implementation.</p>https://resolver.caltech.edu/CaltechETD:etd-05072008-131735Target Tracking Using Clustered Measurements, with Applications to Autonomous Brain-Machine Interfaces
https://resolver.caltech.edu/CaltechETD:etd-05292008-105504
Year: 2008
DOI: 10.7907/6Y3F-8M87
<p>This thesis presents new methods for classifying and tracking the signals of targets that produce clusters of observations, measured in successive recording intervals or scans. This multitarget tracking problem arises, for instance, in extracellular neural recordings, in which an electrode is inserted into the brain to detect the spikes of individual neurons. Since multiple active neurons may lie near the electrode, each detected spike must be assigned to the neuron that produced it, a task known as spike sorting. In the scenario considered in this thesis, the electrode signal is sampled over many brief recording intervals. In each recording interval, all spikes must first be clustered according to their generating neurons, and then each cluster must be associated to clusters from previous recording intervals, thus tracking the signals of putative neuron "targets."</p>
<p>This thesis introduces a novel multitarget tracking solution for the above problem, called multiple hypothesis tracking for clusters (MHTC). The MHTC algorithm has two main parts: a Bayesian clustering algorithm for associating observations to clusters in each interval and a probabilistic supervisory system that manages association hypotheses across intervals. The clustering procedure provides significantly more consistent results than previously available methods, enabling more accurate tracking of targets over time. Such consistency is promoted by a maximum a posteriori (MAP) approach to optimizing a Gaussian mixture model via expectation-maximization (EM), in which information from the preceding intervals serves as a prior for the current interval while still allowing the number and locations of targets to change. MHTC's hypothesis management system, like that of traditional multiple hypothesis tracking (MHT) algorithms, propagates various possibilities for how to assign measurements to existing targets and uses a delayed decision-making logic to resolve data association ambiguities. It also, however, maintains several options, termed model hypotheses, for how to cluster the observations of each interval. This combination of clustering and tracking in a single solution enables MHTC to robustly maintain the identities of cluster-producing targets in challenging recording scenarios.</p>
<p>In addition to these classification and tracking techniques, this thesis presents advances in a miniature robotic electrode microdrive capable of extracellular recordings lasting for days at a time. As a whole, these contributions can play an important role in enabling an autonomous neural interface, which, by frequent automatic repositioning of its recording electrodes, can optimize the recording quality of extracellular signals associated with individual neurons and maintain high quality recordings for long periods of time. Such autonomous movable electrodes may eventually overcome key barriers to engineering a practical neuroprosthetic device and, in the near term, can significantly improve state-of-the-art neuroscience experimental procedures.</p>
https://resolver.caltech.edu/CaltechETD:etd-05292008-105504Continuous Sensorimotor Control Mechanisms in Posterior Parietal Cortex: Forward Model Encoding and Trajectory Decoding
https://resolver.caltech.edu/CaltechETD:etd-05282008-192406
Year: 2008
DOI: 10.7907/VRHH-NT69
<p>During goal-directed movements, primates are able to rapidly and accurately control a movement despite substantial delay times (more than 200 milliseconds) incurred in the sensorimotor control loop. To compensate for these large delays, it has been proposed that the brain uses an internal forward model of the arm to estimate current and upcoming states of a movement, which would be more useful for rapid online control. To study online control mechanisms in the posterior parietal cortex (PPC), we recorded from single neurons while monkeys performed a joystick task. Neurons encoded the static target direction and the dynamic heading direction of the cursor. The temporal encoding properties of many heading neurons reflected a forward estimate of the current state of the cursor that is neither directly available from passive sensory feedback nor compatible with outgoing motor commands, and is thus consistent with PPC serving as a forward model for online sensorimotor control. In addition, we found that the space-time tuning functions of these neurons mostly encode straight and approximately instantaneous trajectories.</p>
<p>Recent advances in cortical prosthetics have focused on recording neural activity in motor cortices and decoding these signals to control the trajectory of a cursor on a computer screen. Building on our encoding results, we demonstrate that joystick-controlled trajectories can also be decoded from PPC ensembles, presumably extracting the dynamic state of the cursor from a forward model. Remarkably, we found that we could accurately reconstruct a monkey’s trajectories using only 5 simultaneously recorded PPC neurons. Furthermore, we tested whether we could decode trajectories during closed-loop brain control sessions, in which the real-time position of the cursor was determined solely by a monkey’s thoughts. The monkey learned to perform brain control trajectories at 80% success rate (for 8 targets) after just 4–5 sessions. This improvement in behavioral performance was accompanied by a corresponding enhancement in neural tuning properties (i.e.,, increased tuning depth and coverage of 2D space) as well as an increase in offline decoding performance of the PPC ensemble. This work marks an important step forward in the development of a neural prosthesis using signals from PPC.</p>
https://resolver.caltech.edu/CaltechETD:etd-05282008-192406Adaptive Feature Selection in Pattern Recognition and Ultra-Wideband Radar Signal Analysis
https://resolver.caltech.edu/CaltechETD:etd-05302008-134607
Year: 2008
DOI: 10.7907/7NR6-AR24
<p>Feature selection from measured data aims to extract informative features to reveal the statistic or stochastic mechanism underlying the complicated or high dimensional original data. In this thesis, the feature selection problem is probed under two situations, one is pattern recognition and the other is ultra-wideband radar signal analysis.</p>
<p>Classical pattern recognition methods select features by their ability to separate the multiple classes with certain gauge measure. The deficiency in this general strategy is its lack of adaptation in specific situations. This deficiency may be overcome by viewing the selected features as a function of not only the training samples but also the unlabeled test data. From this perspective, this thesis proposes an adaptive sequential feature selection algorithm which utilizes an information-theoretic measure to reduce the classification task complexity sequentially, and finally outputs the probabilistic classification result and its variation level. To verify the potential advantage of this algorithm, this thesis applies it to one important problem of neural prosthesis, which concerns decoding a finite number of classes, intended reach directions, from recordings of neural activities in the Parietal Reach Region of one rhesus monkey. Experimental results show that the classification scheme of combining the adaptive sequential feature selection algorithm and the information fusion method outperforms some classical pattern recognition rules, such as the nearest neighbor rule and support vector machine, in decoding performance.</p>
<p>The second scenario in this thesis targets developing a human presence and motion pattern detector through ultra-wideband radar signal analysis. To augment the detection robustness, both static and dynamic features should be utilized. The static features reflect the information of target geometry and its variability, while the dynamic features extract the temporal structure among radar scans. The problem of static feature selection is explored in this thesis, which utilizes the Procrustes shape analysis to generate the representative template for the target images, and makes statistical inference in the tangent space through the Hotelling one sample test. After that, the waveform shape variation structure is decomposed in the tangent space through the principal component analysis. The selected principal components not only accentuate the prominent dynamics of the target motion, but also generate another informative classification feature.</p>
https://resolver.caltech.edu/CaltechETD:etd-05302008-134607Inference in Hybrid Systems with Applications in Neural Prosthetics
https://resolver.caltech.edu/CaltechETD:etd-12312008-184713
Year: 2009
DOI: 10.7907/REB5-BB43
<p>This thesis develops new hybrid system models and associated inference algorithms to create a ``supervisory decoder' for cortical neural prosthetic devices that aim to help the severely handicapped. These devices are a brain-machine interface, consisting of surgically implanted electrode arrays and associated computer decoding algorithms, that enable a human to control external electromechanical devices, such as artificial limbs, by thought alone.</p>
<p>Hybrid systems are characterized by discrete switching between sets of continuous dynamical activity. New hybrid models, which are flexible enough to model neurological activity, are created that incorporate both duration and dynamical state based switching paradigms. Combining generalized linear models with nonstationary and semi-Markov chains gives rise to three new hybrid systems: generalized linear hidden Markov models (GLHMM), hidden semi-Markov models (HSMM) with generalized linear model dynamics, and hidden regressor dependent Markov models (HRDMM). Bayesian inference methods, including variational Bayes and Gibbs sampling, are derived for the identification of existing and developed hybrid models. The developed inference algorithms provide advances over the current hybrid system identification literature by providing a principled way to incorporate prior knowledge and select between alternative model classes and orders, including the number of discrete system states.</p>
<p>Future neuroprostheses that seek to provide a facile interface for the paralyzed patient will require a supervisory decoder that classifies, in real time, the discrete cognitive, behavioral, or planning state of the brain. The developed hybrid models and inference algorithms provide a framework for supervisory decoding, where first, a hybrid-state neurological activity model is identified from data, and then used to estimate the discrete state in real time. The electrical activity of multiple neurons from a cortical area in the brain associated with motor planning (the parietal reach region), and multiple signal types, including both spike arrival times and local field potentials, are fused to give more accurate results. The model structure, including the number of discrete cognitive states, can also be estimated from the data, resulting in significantly improved decoding performance compared to existing methods.</p>
<p>Additional demonstrated applications include the automated segmentation of honey bee motion into discrete primitives, and generating mechanical system models for a pick-and-place machine.</p>
https://resolver.caltech.edu/CaltechETD:etd-12312008-184713Resource Optimization for Networked Estimator with Guaranteed Estimation Quality
https://resolver.caltech.edu/CaltechETD:etd-08272008-121822
Year: 2009
DOI: 10.7907/DTCJ-BN07
Advances in fabrication, modern sensor and communication technologies, and computer architecture have enabled a variety of new networked sensing and control applications. However, many difficulties are inherent with these systems, for example, the constrained communication and computation capabilities, and limited energy resources, which are frequently seen in a wireless sensor network. As a consequence, the networks typically induce many new issues such as limited bandwidth, packet loss, and delay. Estimation and control over such networks thus require new design paradigms beyond traditional sampled-data control, as the aforementioned constraints undoubtedly affect system performance or even stability. In this thesis work, I consider the problem of state estimation over networks. As communication, computation, and energy are scarce resources in such networks, I focus on optimizing the use of them. When the state estimation is carried out over a sensor network, I consider the problem of minimizing the sensor energy usage and maximizing the network lifetime. When the state estimation is carried out over a packet-delaying network, I consider the problem of minimizing the buffer length at the remote state estimator. In each scenario, a certain desired level of estimation quality is guaranteed.
https://resolver.caltech.edu/CaltechETD:etd-08272008-121822Neural Control and Biomechanics of Flight Initiation in Drosophila melanogaster
https://resolver.caltech.edu/CaltechETD:etd-05282009-215548
Year: 2009
DOI: 10.7907/PR7S-Y618
In response to abrupt visual stimulation, the fruit fly, Drosophila melanogaster, quickly initiates flight. This rapid takeoff is believed to be a reflex coordinated by a pair of large descending interneurons (the "giant fibers"). However, it has been difficult to evoke escapes in wild-type flies, and thus flight initiation behavior in the unrestrained wild-type fly is poorly described. I have taken advantage of recent advances in high-speed videography to capture video sequences of Drosophila flight initiation at the temporal resolution of 6,000 frames per second. A three-dimensional kinematic analysis of takeoff sequences indicates that wing use during the jumping phase of flight initiation is essential for stabilizing flight. During voluntary takeoffs, flies raise their wings prior to leaving the ground, resulting in a stable, controlled takeoff. In contrast, during visually-elicited escapes flies pull their wings down close to their body during the takeoff jump, resulting in tumbling flights that are faster but less steady. The takeoff kinematics suggest that the power delivered by the legs is substantially greater during these escapes than during voluntary takeoffs. Thus, I show that the two types of Drosophila flight initiation result in different flight performances once the fly is airborne, and that these performances are distinguished by a trade-off between speed and stability. I also determined that flies can use visual information to plan a jump directly away from a looming threat. This is surprising, given the simple architecture of the giant fiber pathway thought to mediate escape. I found that approximately 200 ms before takeoff, flies begin a series of postural adjustments that determine the direction of their escape. These movements position their center of mass so that leg extension will push them away from the looming stimulus. These preflight movements are not the result of a simple feed-forward motor program because their magnitude and direction depend on the flies' initial postural state. Furthermore, flies plan a takeoff direction even in instances when they choose not to jump. This sophisticated motor program is evidence for a form of rapid, visually mediated motor planning in a genetically accessible model organism.
https://resolver.caltech.edu/CaltechETD:etd-05282009-215548Tracker Effector-Specific and Motor Planning Signals in Human Frontal and Parietal Cortices: Relevance for Goal-Directed Action and Neural Prosthetics
https://resolver.caltech.edu/CaltechETD:etd-10302008-224422
Year: 2009
DOI: 10.7907/CDC5-TK34
Delayed response tasks and functional magnetic resonance imaging were employed to map the neural architecture underlying goal-directed action planning in the human brain, examine interactions between motor planning and effector-specification (arm vs. eye), and explore other related processes and variables. Studies in healthy human subjects revealed a frontoparietal network of brain areas selectively involved in motor planning compared to control processes. Nodes within this network were characterized based on their functional properties, including effector-specificity. In frontal cortex, the dorsal premotor and supplementary motor areas preferentially encoded motor plans for arm reaches compared to saccadic eye movements, while the inferior frontal eye field was identified based on its selective involvement in eye movements. In parietal cortex, a similar dissociation of arm- and eye-specific brain areas was observed in the superior lobule. A medial branch of the intraparietal sulcus preferentially encoded eye movements, in contrast to more anterior medial, and posterior medial, portions of the intraparietal sulcus that preferentially encoded arm movements. Additionally, motor planning areas were engaged during voluntary shifts of spatial attention and during working memory for visual cues when these cues were relevant for upcoming movements. Many of these brain areas also encoded the type of arm movement (reach vs. point), arm posture, and limb contralaterality, a property that co-varied with increasing ties to motor execution. Also, a comparison of real vs. imagined arm movements revealed that the imagined arm could be used as a proxy for the real arm to drive activity in motor planning areas. Another study completed in healthy control and spinal cord-injured subjects demonstrated the preservation of a relatively normal pattern of brain activity after the brain is functionally disconnected from the limbs. The degree of preservation of healthy/normal BOLD activity levels, particularly in the medial parietal cortex, strongly correlated with clinical and behavioral variables and could predict functional motor improvements in spinal cord-injured subjects six months later. These studies contribute to our understanding of the representation of goal-directed action planning in the human brain, elucidate human-monkey interspecies functional homologies, and have implications for the design and implantation of cortical neural prosthetic devices.
https://resolver.caltech.edu/CaltechETD:etd-10302008-224422Stochastic Analysis, Model and Reliability Updating of Complex Systems with Applications to Structural Dynamics
https://resolver.caltech.edu/CaltechETD:etd-05292009-102458
Year: 2009
DOI: 10.7907/K5T7-4B72
<p>In many engineering applications, it is a formidable task to construct mathematical models that are expected to produce accurate predictions of the behavior of a system of interest. During the construction of such predictive models, errors due to imperfect modeling and uncertainties due to incomplete information about the system and its environment (e.g., input or excitation) always exist and can be accounted for appropriately by using probability logic. To assess the system performance subjected to dynamic excitations, a stochastic system analysis considering all the uncertainties involved has to be performed. In engineering, evaluating the robust failure probability (or its complement, robust reliability) of the system is a very important part of such stochastic system analysis. The word ‘robust’ is used because all uncertainties, including those due to modeling of the system, are taken into account during the system analysis, while the word ‘failure’ is used to refer to unacceptable behavior or unsatisfactory performance of the system output(s). Whenever possible, the system (or subsystem) output (or maybe input as well) should be measured to update models for the system so that a more robust evaluation of the system performance can be obtained. In this thesis, the focus is on stochastic system analysis, model and reliability updating of complex systems, with special attention to complex dynamic systems which can have high-dimensional uncertainties, which are known to be a very challenging problem. Here, full Bayesian model updating approach is adopted to provide a robust and rigorous framework for these applications due to its ability to characterize modeling uncertainties associated with the underlying system and to its exclusive foundation on the probability axioms.</p>
<p>First, model updating of a complex system which can have high-dimensional uncertainties within a stochastic system model class is considered. To solve the challenging computational problems, stochastic simulation methods, which are reliable and robust to problem complexity, are proposed. The Hybrid Monte Carlo method is investigated and it is shown how this method can be used to solve Bayesian model updating problems of complex dynamic systems involving high-dimensional uncertainties. New formulae for Markov Chain convergence assessment are derived. Advanced hybrid Markov Chain Monte Carlo simulation algorithms are also presented in the end.</p>
<p>Next, the problem of how to select the most plausible model class from a set of competing candidate model classes for the system and how to obtain robust predictions from these model classes rigorously, based on data, is considered. To tackle this problem, Bayesian model class selection and averaging may be used, which is based on the posterior probability of different candidate classes for a system. However, these require calculation of the evidence of the model class based on the system data, which requires the computation of a multi-dimensional integral involving the product of the likelihood and prior defined by the model class. Methods for solving the computationally challenging problem of evidence calculation are reviewed and new methods using posterior samples are presented. </p>
<p>Multiple stochastic model classes can be created even there is only one embedded deterministic model. These model classes can be viewed as a generalization of the stochastic models considered in Kalman filtering to include uncertainties in the parameters characterizing the stochastic models. State-of-the-art algorithms are used to solve the challenging computational problems resulting from these extended model classes. Bayesian model class selection is used to evaluate the posterior probability of an extended model classe and the original one to allow a data-based comparison. The problem of calculating robust system reliability is also addressed. The importance and effectiveness of the proposed method is illustrated with examples for robust reliability updating of structural systems. Another significance of this work is to show the sensitivity of the results of stochastic analysis, especially the robust system reliability, to how the uncertainties are handled, which is often ignored in past studies.</p>
<p>A model validation problem is then considered where a series of experiments are conducted that involve collecting data from successively more complex subsystems and these data are to be used to predict the response of a related more complex system. A novel methodology based on Bayesian updating of hierarchical stochastic system model classes using such experimental data is proposed for uncertainty quantification and propagation, model validation, and robust prediction of the response of the target system. Recently-developed stochastic simulation methods are used to solve the computational problems involved.</p>
<p>Finally, a novel approach based on stochastic simulation methods is developed using current system data, to update the robust failure probability of a dynamic system which will be subjected to future uncertain dynamic excitations. Another problem of interest is to calculate the robust failure probability of a dynamic system during the time when the system is subjected to dynamic excitation, based on real-time measurements of some output from the system (with or without corresponding input data) and allowing for modeling uncertainties; this generalizes Kalman filtering to uncertain nonlinear dynamic systems. For this purpose, a novel approach is introduced based on stochastic simulation methods to update the reliability of a nonlinear dynamic system, potentially in real time if the calculations can be performed fast enough.</p>https://resolver.caltech.edu/CaltechETD:etd-05292009-102458Safety Verification and Failure Analysis of Goal-Based Hybrid Control Systems
https://resolver.caltech.edu/CaltechETD:etd-05292009-111937
Year: 2009
DOI: 10.7907/3H42-BF56
The success of complex autonomous robotic systems depends on the quality and correctness of their fault tolerant control systems. A goal-based approach to fault tolerant control, which is modeled after a software architecture developed at the Jet Propulsion Laboratory, uses networks of goals to control autonomous systems. The complex conditional branching of the control program makes safety verification necessary. Three novel verification methods are presented. In the first, goal networks are converted to linear hybrid automata via a bisimulation. The converted automata can then be verified against an unsafe set of conditions using an existing symbolic model checker such as PHAVer. Due to the complexity issues that result from this method, a design for verification software tool, the SBT Checker, was developed to create goal networks that have state-based transitions. Goal networks that have state-based transitions can be converted to hybrid automata whose locations' invariants contain all information necessary to determine the transitions between the locations. An original verification software called InVeriant can then be used to find unsafe locations of linear hybrid systems based on the locations’ invariants and rate conditions, which are compared to the unsafe set of conditions. The reachability of the unsafe locations depends only on the reachability of the states of the state variables constrained in the locations' invariants from those state variables' initial conditions. In cases where this reachability condition is not trivially true, the software efficiently searches for a path to the unsafe locations using properties of the system. The third verification method is the calculation of the failure probability of the verified hybrid control system due to state estimation uncertainty, which is extremely important in autonomous systems that rely heavily on the state estimates made from sensor measurements. Finally, two significant example goal network control programs, one for a complex rover and another for a proposed aerobot mission to Titan, a moon of Saturn, are verified using the three techniques presented.https://resolver.caltech.edu/CaltechETD:etd-05292009-111937Integrated Retinal Implants
https://resolver.caltech.edu/CaltechETD:etd-02162009-095558
Year: 2009
DOI: 10.7907/AMK6-TA42
<p>Integrated wireless implants have always been the ultimate goal for neural prostheses. However, technologies are still in development and few have actually been transferred to clinical practice due to constraints in material biocompatibility, device miniaturization and flexibility. In this dissertation, emphasis is placed on the development of Parylene devices for neural prostheses, and particularly, for retinal prostheses that partially restore lost vision for patients suffering from outer retina degeneration.</p>
<p>A basic Parylene-metal-Parylene skin technology for making planar Parylene micro-electro-mechanical systems (MEMS) devices, such as electrode arrays and radio-frequency (RF) coil, is first discussed, followed by accelerated lifetime soaking tests to investigate the long term stability of such skins in hot saline under both passive and active electrical stressing. Discussion is further expanded on a detailed description of the design, fabrication, and testing procedure of two types of MEMS coils, which serve as receiver coils for wireless power and data transfer in a retinal implant system. After that, an embedded chip integration technology is presented, which allows the integration of complementary metal-oxide-semiconductor (CMOS) integrated circuit (IC) chips with other MEMS devices and discrete components so as to achieve high-level system functionality. Finally, an integrated wireless neural stimulator is designed and successfully fabricated using a test chip.</p>
https://resolver.caltech.edu/CaltechETD:etd-02162009-095558Egocentric Distance Encoding in the Posterior Parietal Cortex
https://resolver.caltech.edu/CaltechETD:etd-12062008-112215
Year: 2009
DOI: 10.7907/FBNQ-0P48
<p>Previous studies have shown that the parietal reach region (PRR) encodes the two dimensional location of frontoparallel reach targets in an eye centered reference frame in early movement plans (Batista 1999; Snyder et al. 2000). Performing a visually guided reach initially requires the ability to perceive the depth of a target in three dimensional space. Beyond that initial perception, however, reach plans may represent the egocentric distance of the target in different ways. To investigate how a reach target is represented in three dimensions, recordings were made of the spiking activity of PRR neurons from two rhesus macaques trained to fixate and perform memory reaches to targets at different depths. Reach and fixation targets were configured to explore whether neural activity directly reflects egocentric distance as the amplitude of the required motor command, which is the absolute depth of the target, or rather the relative depth of the target with reference to fixation depth.</p>
<p>This thesis shows that planning activity in PRR represents the depth of the reach target as a function of disparity and fixation depth, the spatial parameters important for encoding the egocentric distance of a reach goal in an eye centered reference frame. Most PRR neurons were found to be sensitive to the disparity of a reach target (82%), and vergence angle (74%) which determines fixation depth. Most PRR neurons carry both disparity and vergence angle signals, and comparisons of several modulation indices and the information carried in firing rates for each variable show a single homogenous PRR population encodes egocentric distance. The strength of modulation by disparity was maintained across vergence angle, where vergence angle gain modulates disparity tuning while preserving the location of peak tuning features in PRR neurons, which allows the absolute depth of the reach target to be decoded from the population activity. Neural activity in PRR shows a wide range of sensitivity to both target disparity and fixation depth that has never been previously tested in a reach task. The results demonstrate a specific role for PRR neurons in supporting eye-hand coordination when decoupling the effector from the point of gaze.</p>
https://resolver.caltech.edu/CaltechETD:etd-12062008-112215Dynamic Simulation and Control of Articulated Limbs
https://resolver.caltech.edu/CaltechETD:etd-02162009-054828
Year: 2009
DOI: 10.7907/2FAM-6A26
Many useful mechanisms can be modelled as articulated systems: collections of rigid bodies linked together with joints that constrain relative movement. The two parts of this thesis study the complementary problems of simulation and control for such systems. In the first part, we describe an implementation and extension of a physically based modelling framework known as "dynamic constraints" in which forces of constraint linking bodies in an articulated system are explicitly calculated. In addition to identifying some important robustness and stability issues for these calculations, we extend the framework to systems whose internal degrees of freedom can be directly parameterized. This permits significant efficiency gains for mechanisms which model limbs. The second part of the thesis centers on the adaptive control of limb configuration through simulated actuators. In this problem, the nonlinear structure and parametric details of a limb are assumed to be unknown. We present and illustrate the performance of an adaptive scheme which performs considerably better than conventional nonadaptive techniques, and which is competitive with adaptive methods which use more a priori knowledge of limb dynamics. https://resolver.caltech.edu/CaltechETD:etd-02162009-054828Implantable Wireless Intraocular Pressure Sensors
https://resolver.caltech.edu/CaltechETD:etd-09022008-113511
Year: 2009
DOI: 10.7907/46T7-0P24
<p>The work in this thesis aims to develop a suite of biomedical microdevice implants, with an intense focus on pressure sensors, for glaucoma study and management featuring the enabling micro-electro-mechanical-system (MEMS) technologies and the use of parylene (poly-para-xylene) as a biocompatible MEMS material. The problems of the debilitating eye disease glaucoma threaten tens of millions of people worldwide with loss of vision, and are not completely resolved using the current non-optimal clinical procedures. Given the relation of neuropathy and the physiological parameter of intraocular pressure (IOP) in glaucoma from clinical findings, such parylene-based MEMS implants are investigated to realize physical IOP monitoring and regulation, and further to accomplish continuous, direct, accurate, reliable, and more effective glaucoma detection and treatment.</p>
<p>Miniaturized parylene-based passive pressure sensors are presented in this thesis for IOP monitoring. Complete design, fabrication, characterization, and analysis of such MEMS implants are described to demonstrate their feasibility, covering both engineering and surgical/biological aspects of the proposed applications. Their passive behaviors, based on the comprised micromechanical structures, facilitate unpowered device operations. In addition, such devices are microfabricated in suitable form factors so that minimally invasive suture-less implantation procedures are possible, minimizing time and complexity of the surgeries. Two types of micromachined wireless pressure sensors are developed utilizing optical and electrical sensing methodologies, respectively, to explore the possibility of the proposed implant approach. On-bench experimental results verify that wireless pressure sensing with 1 mmHg accuracy in the 0–100 mmHg range can be achieved using both types of devices. Surgical studies, including ex vivo and in vivo animal tests, confirm the bioefficacy and biostability of the device implants in the intraocular environment. With the attempt of providing implementation concepts of the MEMS implant approaches for ultimate glaucoma study and management in practice, system-level designs and configurations involving such microdevice implants are briefly described as well. Micromachined passive-valved flow-control devices with designed surgical and engineering features are also developed (experimentally achieving 0–100 mmHg and 0–10 uL/min pressure and flow rate regulation ranges) to investigate the feasibility and possibility of such implant approach for unpowered physical IOP regulation in glaucoma treatment.</p>
https://resolver.caltech.edu/CaltechETD:etd-09022008-113511Variational Methods for Control and Design of Bipedal Robot Models
https://resolver.caltech.edu/CaltechTHESIS:05282010-094801935
Year: 2010
DOI: 10.7907/KATX-3233
<p>This thesis investigates nonsmooth mechanics using variational methods for the modeling, control, and design of bipedal robots.</p>
<p>The theory of Lagrangian mechanics is extended to capture a variety of nonsmooth collision behaviors in rigid body systems. Notably, a variational impact model is presented for the transition of constraints behavior that describes a biped switching stance feet at the conclusion of a step.</p>
<p>Next, discretizations of the impact mechanics are developed using the framework of variational discrete mechanics. The resulting variational collision integrators are consistent with the continuous time theory and have an underlying symplectic structure.</p>
<p>In addition to their role as integrators, the discrete equations of motion capturing nonsmooth dynamics enable a direct method for trajectory optimization. Upon specifically defining the optimal control problem for nonsmooth systems, examples demonstrate this optimization method in the task of determining periodic gaits for
two rigid body biped models.</p>
<p>An additional effort is made to optimize bipedal walking motions through modifications in system design. A method for determining optimal designs using a combination
of trajectory optimization methods and surrogate function optimization methods is defined. This method is demonstrated in the task of determining knee joint placement
in a given biped model.</p>https://resolver.caltech.edu/CaltechTHESIS:05282010-094801935A Parylene Real Time PCR Microdevice
https://resolver.caltech.edu/CaltechTHESIS:12142009-145646250
Year: 2010
DOI: 10.7907/YC9S-0R15
<p>The polymerase chain reaction (PCR) is a powerful biochemical assay that is used in virtually all biochemical labs. By specifically amplifying a small sample of DNA, this technique is useful in the fields of paternity testing, forensics, and virus detection, just to name a few. A useful advancement of PCR involves monitoring the fluorescence generated by an increase in DNA during the amplification. This so called real time (RT)PCR allows quantification of the initial sample amount and allows for shorter assay times by stopping the reaction when enough fluorescence has been detected.</p>
<p>Technology in the field of micro-electro-mechanical systems (MEMS) has advanced from the academic laboratory level to a handful of commercially successful devices. Work on adapting MEMS to biochemical applications, however, is still at the laboratory research stage. Recent breakthroughs in the use of more biocompatible materials in MEMS devices have helped to advance bio-MEMS. In particular, the polymer Parylene has superior properties that present a promising new platform for this field.</p>
<p>This work presents the design, fabrication, and testing of a parylene-based MEMS RTPCR device. By combining advancements in both biology and MEMS engineering, this work demonstrates the feasibility of such a device along with quantitative analysis and data that serve as a guide for its future development.</p>
https://resolver.caltech.edu/CaltechTHESIS:12142009-145646250Real-Time Applications of 3D Object Detection and Tracking
https://resolver.caltech.edu/CaltechTHESIS:01152010-143831008
Year: 2010
DOI: 10.7907/4N1K-GK74
Robot perception is a fundamental aspect of any autonomous system. It gives the robot the capacity to make sense of vast amounts of data and gain an understanding of the world around it. An active problem in the area of robot perception is real-time detection and pose estimation of 3D objects. This thesis presents an approach to 3D object detection and tracking utilizing a stereo-camera sensor. Geometric object models are learned in short order time via a training phase and real-time detection and tracking is made possible by performing sparse stereo calculations on the chosen features within an adaptive region of interest of the camera image. The experimental results obtained by using this method will show the effectiveness of the approach as compared against ground truth measures in real-time. Using that framework as a basis, extensions to two other problems in robot sensing are then considered: (1) sensor-planning for model identification, and (2) sensor-planning for object-search. In the former, a novel algorithm for determining the next-best-view for a mobile sensor to identify an unknown 3D object from among a database of known models is presented and tested across two experiments involving real robotic systems. An information theoretic approach is taken to quantify the utility of each potential sensing action and the validity of the algorithm is discussed. In the latter area, a novel approach is presented that allows an autonomous mobile robot to search for a 3D object using an onboard stereo camera sensor mounted on a pan-tilt head. Search efficiency is realized by the combination of a coarse-scale global search coupled with a fine-scale local search, guided by a grid-based probability map. Obstacle avoidance during the search is naturally integrated into the method with additional experimental results on a mobile robot presented to illustrate and validate the proposed search strategy. All presented experiments were carried out in real-time processing with modest computation done by a single laptop computer.https://resolver.caltech.edu/CaltechTHESIS:01152010-143831008Robot Motion Planning in Dynamic, Cluttered, and Uncertain Environments: the Partially Closed-Loop Receding Horizon Control Approach
https://resolver.caltech.edu/CaltechTHESIS:02042010-152638957
Year: 2010
DOI: 10.7907/SD3N-JR18
This thesis is concerned with robot motion planning in dynamic, cluttered, and uncertain environments. Successful and efficient robot operation in such environments requires reasoning about the future system evolution and the uncertainty associated with obstacles and moving agents in the environment. Current motion planning strategies ignore future information and are limited by the resulting growth of uncertainty as the system is evolved. This thesis presents an approach that accounts for future information gathering (and the quality of that information) in the planning process. The Partially Closed-Loop Receding Horizon Control approach, introduced in this thesis, is based on Dynamic Programming with imperfect state information. Probabilistic collision constraints, due to the need for obstacle avoidance between the robot and obstacles with uncertain locations and geometries, are developed and imposed. By accounting for the anticipated future information, the uncertainty associated with the system evolution is managed, allowing for greater numbers of moving agents and more complex agent behaviors to be handled. Simulation results demonstrate the benefit of the proposed approach over existing approaches in static and dynamic environments. Complex agent behaviors, including multimodal and interactive agent-robot models, are considered.https://resolver.caltech.edu/CaltechTHESIS:02042010-152638957Formal Methods for Design and Verification of Embedded Control Systems: Application to an Autonomous Vehicle
https://resolver.caltech.edu/CaltechTHESIS:05272010-153304667
Year: 2010
DOI: 10.7907/XZ3X-7V51
<p>The design of reliable embedded control systems inherits the difficulties involved in designing both control systems and distributed (concurrent) computing systems. Design bugs in these systems may arise from the unforeseen interactions among the computing, communication and control subsystems. Motivated by the difficulties of finding this type of design bugs, this thesis develops mathematical frameworks, based on formal methods, to facilitate the design and analysis of such embedded systems. An expressive specification language of linear temporal logic (LTL) is used to specify the desired system properties. The practicality of the proposed frameworks is demonstrated through autonomous vehicle case studies and autonomous urban driving problems.</p>
<p>Our approach incorporates methodology from computer science and control, including model checking, theorem proving, synthesis of digital designs, reachability analysis, Lyapunov-type methods and receding horizon control. This thesis consists of two complementary parts, namely, verification and design. First, we introduce Periodically Controlled Hybrid Automata (PCHA), a subclass of hybrid automata that abstractly captures a common design pattern in embedded control systems. New sufficient conditions that exploit the structure of PCHAs in order to simplify their invariant verification are presented.</p>
<p>Although the aforementioned technique simplifies an invariant verification of PCHAs, finding a proper invariant remains a challenging problem. To complement the verification efforts, in the second part of the thesis, we present a methodology for automatic synthesis of embedded control software that provides a formal guarantee of system correctness, with respect to its desired properties expressed in linear temporal logic. The correctness of the system is guaranteed even in the presence of an adversary (typically arising from changes in the environments), disturbances and modeling errors. A receding horizon framework is proposed to alleviate the associated computational complexity of LTL synthesis. The effectiveness of this framework is demonstrated through the autonomous urban driving problems.</p>
https://resolver.caltech.edu/CaltechTHESIS:05272010-153304667Neuro-Evolution Using Recombinational Algorithms and Embryogenesis for Robotic Control
https://resolver.caltech.edu/CaltechTHESIS:06092010-140839602
Year: 2010
DOI: 10.7907/YNED-VN66
Control tasks involving dramatic nonlinearities, such as decision making, can be challenging for classical design methods. However, autonomous, stochastic design methods such as evolutionary computation have proved effective. In particular, genetic algorithms that create designs via the application of recombinational rules are robust and highly scalable. Neuro-Evolution Using Recombinational Algorithms and Embryogenesis (NEURAE) is a genetic algorithm that creates C++ programs that in turn create neural networks which can function as logic gates. The neural networks created are scalable and robust enough to feature redundancies that allow the network to function despite internal failures. An analysis of NEURAE evinces how biologically inspired phenomena apply to simulated evolution. This allows for an optimization of NEURAE that enables it to create controllers for a simulated swarm of Khepera-inspired robots.https://resolver.caltech.edu/CaltechTHESIS:06092010-140839602Integrated Microfluidic Devices for Cell Culture and Assay
https://resolver.caltech.edu/CaltechTHESIS:10162009-053129363
Year: 2010
DOI: 10.7907/D43B-D825
<p>This thesis presents the development of three-dimensional (3-D) microfluidic devices for cellular studies, with focus on applications for high-throughput cell culture and cell-based assay. Microfluidic devices provide potential inexpensive platforms for high-throughput screening with the advantages of precise liquid handling, ability to control cell culture microenvironment, and reduced reagents and cells.</p>
<p>Because a mixture of drugs or chemical compounds can often treat diseases more effectively or act synergistically in certain cellular pathways, a device capable of screening the combinatorial effects of multiple compound exposures on cells is highly desirable. To this end, a novel method to monolithically fabricate 3-D microfluidic networks was developed, and based on this fabrication technology, the first cell culture device with an integrated combinatorial mixer was constructed. The proof-of-concept chip having a three-input combinatorial mixer and eight individually isolated micro culture chambers was fabricated on silicon utilizing the surface micromachining of Parylene C (poly(chloro-p-xylylene)). Unlike other 3-D microfluidic fabrications, multilayer bonding process was favorably obviated. By incorporating several microfluidic overpass structures to allow one microfluidic channel to cross over other microfluidic channels, the combinatorial mixer generated all the combinations of the input fluidic streams. Cell culturing on-chip was successful, and the ability to simultaneously treat arrays of cells with different combinations of compounds was demonstrated.</p>
<p>To facilitate cell-based assay, another combinatorial cell array device was fabricated on glass with incorporated membrane. Characterization of the combined compound concentration profile at each chamber with a fluorescence method was developed. We demonstrated functionality of the quantitative cell-based assay by screening three different compounds’ ability to reduce cytotoxicity of hydrogen peroxide on neuron cells and also assaying combinatorial exposures of three chemotherapeutic agents on breast cancer cells. The 3-D microfluidic fabrication process was extended to construct multilayer microfluidic device with integrated membrane. Applications of microfluidic devices for marine microbiology were demonstrated. Based on the capabilities demonstrated in this work, devices with high-density cell array and integrated high-input combinatorial mixer can be constructed. At the same time, the technology has general applicability for building complex 3-D microfluidic devices, which can broaden the applications for current lab-on-a-chip systems.</p>
https://resolver.caltech.edu/CaltechTHESIS:10162009-053129363Searching Large-Scale Image Collections
https://resolver.caltech.edu/CaltechTHESIS:04252011-145432540
Year: 2011
DOI: 10.7907/VRGJ-4J54
Searching quickly and accurately in a large collection of images has become an increasingly important problem. The ultimate goal is to make visual search possible: allow users to search using images in addition to typing text. The typical approach is to index all the images of interest (e.g., images of landmarks, books, or DVDs) in a database and let users question the system with query images. Such a database can reach billions of images, and this poses challenges in terms of memory and computational requirements and recognition performance. In this work we provide an in depth study of systems used for searching large-scale image collections.
Specifically, we provide a thorough comparison of the two leading image search approaches: Full Representation (FR) vs. Bag of Words (BoW). We derive theoretical estimates of how the memory and computational cost scale with the number of images in the database, and empirically evaluate the performance and run time on four real-world datasets. Our experiments suggest that FR provides better recognition performance than BoW, though it requires more memory. Therefore, we address these shortcomings by presenting novel methods that increase the recognition performance of BoW and decrease the memory requirements of FR. Finally, we present a novel way to parallelize FR on multiple machines and scale up database sizes to 100 million images with interactive run time.https://resolver.caltech.edu/CaltechTHESIS:04252011-145432540Steady as She Goes: Visual Autocorrelators and Antenna-Mediated Airspeed Feedback in the Control of Flight Dynamics in Fruit Flies and Robotics
https://resolver.caltech.edu/CaltechThesis:06082011-191034348
Year: 2011
DOI: 10.7907/Z3D0-GG27
Achieving agile autonomous flight by an insect-sized micro aerial vehicle (MAV) will require improved technology that is radically smaller, lighter, and more power-efficient. One animal that has solved the problem is the fly, a virtuoso among insect flyers whose nervous system can perform sophisticated aerial maneuvers under severe computational constraints. This thesis is concerned with understanding and emulating the dynamics of the fly's feedback control system. Because vision is noisy and information rich, processing time may a problem for a fast-moving MAV or fly. By tracking the fruit fly Drosophila melanogaster in free flight in gusts of wind, I found that they incorporate feedback from wind-sensing antennae in a fast feedback loop that dampens the forward-flight dynamics. The slower dynamics are easier to control for long-delay visual feedback, making the fly more robust to the limitations of its visual system. Using the fly as inspiration, I designed a minimal, visual autocorrelation based controller that used a small array of visual sensors to stabilize a fan-actuated hovercraft robot in a narrow corridor. Using a model for correlators developed for the robot, I showed that a uniform array of visual correlators was sufficient to explain the free-flight velocity regulation behavior of flies, rather than a different model. In addition to illustrating the benefits of concurrent scientific analysis and engineering synthesis, the results give new insight into how to control small biological and man-made flying vehicles using limited, noisy sensors.https://resolver.caltech.edu/CaltechThesis:06082011-191034348MEMS Electrolytic Inchworms for Movable Neural Probe Applications
https://resolver.caltech.edu/CaltechTHESIS:01102011-204907918
Year: 2011
DOI: 10.7907/GQYY-BM80
<p>Over decades of cortical neural prosthesis, it was found that "movable" neural probes are important to track neurons for long-term, reliable prostheses. This is challenging because the ideal movable probes require low voltage, small power, bidirectional/latchable movement, and large total traveling distance. The device should also be small enough to entirely fit under the skull after implantation. Many different devices have been demonstrated to move neural probes, but none of them satisfies all the actuation and size requirements.</p>
<p>This thesis presents our work on actuators for movable neural probes that combine MEMS technology with an electrolytic actuation mechanism. Each inchworm is based on two electrolytic balloon actuators. The actuators rely on gas generation by electrolysis inside a sealed balloon, which causes its expansion. When electrolysis is stopped, gas recombination and permeation across the balloon membrane cause the balloon to relax. Electrolytic actuation, although slow, has several advantages: low power, low voltage, and ability to provide large force and displacement. The balloons have been characterized and their behavior mathematically modeled. Innovative salt-shell-based and hydrogel-based processes have been developed to fabricate the balloons and to allow their replenishment by osmosis.</p>
<p>Two balloons are combined into a bidirectional inchworm mechanism. Large traveling distance can be obtained in multiple cycles, the only constraint being the probe length. Displacement of a silicon probe and of a commercial metal probe have been demonstrated in both directions, with a displacement per cycle between 0.5 um and 75 um. The voltage required to drive electrolysis is typically around 3.5 V, with peak power per balloon around 100 uW. The devices were tested in air, water, and saline.</p>
<p>Closed-loop control of the inchworm may be needed for accurate positioning of the probe, and monitoring of the pressure inside the balloons represents a possible source of feedback from the inchworm. Parylene-membrane pressure sensors that are suitable for integration inside balloon actuators have been demonstrated.</p>https://resolver.caltech.edu/CaltechTHESIS:01102011-204907918Towards Open Ended Learning: Budgets, Model Selection, and Representation
https://resolver.caltech.edu/CaltechTHESIS:02092011-171146758
Year: 2011
DOI: 10.7907/T92X-DQ05
<p>Biological organisms learn to recognize visual categories continuously over the course of their lifetimes. This impressive capability allows them to adapt to new circumstances as they arise, and to flexibly incorporate new object categories as they are discovered. Inspired by this capability, we seek to create artificial recognition systems that can learn in a similar fashion.</p>
<p>We identify a number of characteristics that define this Open Ended learning capability. Open Ended learning is unsupervised: object instances need not be explicitly labeled with a category indicator during training. Learning occurs incrementally as experience ensues; there is no training period that is distinct from operation and the categorization system must operate and update itself in a timely fashion with limited computational resources. Open Ended learning systems must flexibly adapt the number of categories as new evidence is uncovered.</p>
<p>Having identified these requirements, we develop Open Ended categorization systems based on probabilistic graphical models and study their properties. From the perspective of building practical systems, the most challenging requirement of Open Ended learning is that it must be carried out in an unsupervised fashion. We then study the question of how best to represent data items and categories in unsupervised learning algorithms in order to extend their domain of application.</p>
<p>Finally, we conclude that continuously learning categorization systems are likely to require human intervention and supervision for some time to come, which suggests research in how best to structure machine-human interactions. We end this thesis by studying a system that reverses the typical role of human and machine in most learning systems. In Crowd Clustering, humans perform the fundamental image categorization tasks, and the machine learning system evaluates and aggregates the results of human workers.</p> https://resolver.caltech.edu/CaltechTHESIS:02092011-171146758Flexible Neural Implants
https://resolver.caltech.edu/CaltechTHESIS:07092010-104142755
Year: 2011
DOI: 10.7907/VMZB-0N20
<p>Despite recent development in integration technologies for biomedical implantable devices, current state-of-the-art prosthetic platforms still lack a reliable and convenient packaging scheme to integrate high-density signal-driving chips, wireless telemetry circuitries and noise-canceling amplifiers, mainly due to the limitations in fabrication technology, material compatibility and interconnect reliability. In this dissertation, new packaging technologies are developed and presented to enable a new generation of flexible neural implants. These technologies can also house integrated circuit chips and provide high-density electrical connection to it.</p>
<p>This packaging scheme utilizes the parylene-metal-parylene skin structure and can be totally integrated and be monolithically fabricated with existing functional devices. The size and the electrode patterns can be modified to suit different chips and applications. Integration with flexible cable integrated silicon probes for neural prosthesis, implantable muscle stimulators and implantable RFID tagging technology are all successfully demonstrated in this dissertation. Other discrete components can also be integrated to achieve high level functionality.</p>
<p>In order to ensure the long-term stability of such packaging scheme, accelerated hot saline soaking test is conducted on the overall structure and its components. Detailed adhesion enhancement techniques are also presented to improve its performances. A physical model of the flexible retinal implant is then tested in vivo during the course of the experiment. Finally, the high-density squeegee bonding technique is introduced, which allows the integration of a 256-channel chip. Functionality of the chip has been demonstrated. As a result, this technology has the potential to achieve ultra high lead count connection and can facilitate future research in flexible implantable biodevices.</p>
https://resolver.caltech.edu/CaltechTHESIS:07092010-104142755Network Structure Optimization with Applications to Minimizing Variance and Crosstalk
https://resolver.caltech.edu/CaltechTHESIS:12102011-161913831
Year: 2012
DOI: 10.7907/ER8Y-ZK49
This thesis provides a unified methodology for analyzing structural properties of graphs, along with their applications. In the last several years, the field of complex networks has been extensively studied, and it is now well understood that the way a large network is built is closely intertwined with its function. Structural properties have an impact on the function of the network, and the form of many systems has been evolved in order to optimize for given functions. Despite the great progress, particularly in how structural attributes affect the various network functions, there is a significant gap in the quantitative study of how much these properties can change in a network without a significant impact on the functionality of the system, or what the bounds of these structural attributes are. Here, we find and analytically prove tight bounds of global graph properties, as well as the form of the graphs that achieve these bounds. The attributes studied include the network efficiency, radius, diameter, average distance, betweenness centrality, resistance distance, and average clustering. All of these qualities have a direct impact on the function of the network, and finding the graph that optimizes one or more of them is of interest when designing a large system. In addition, we measure how sensitive these properties are with respect to random rewirings or addition of new edges, since designing a network with a given set of constraints may include a lot of trade-offs. This thesis also studies properties that are of interest in both natural and engineered networks, such as maximum immunity to crosstalk interactions and random noise. We are primarily focused on networks where information is transmitted through a means that is accessible by all the individual units of the network and the interactions among the different entities that comprise it do not necessarily have a dedicated mechanism that facilitates information transmission, or isolates them from other parts of the network. Two examples of this class are biological and chemical reaction networks. Such networks suffer from unwanted crosstalk interactions when two or more units spuriously interact with each other. In addition, they are subject to random fluctuations in their output, both due to noisy inputs and because of the random variance of their parameters. These two types of randomness affect the behavior of the system in ways that are intrinsically different. We examine the network topologies that accentuate or alleviate the effect of random variance in the network for both directed and undirected graphs, and find that increasing the crosstalk among different parts reduces the output variance but also contributes to a slower response.
https://resolver.caltech.edu/CaltechTHESIS:12102011-161913831Axel Rover Tethered Dynamics and Motion Planning on Extreme Planetary Terrain
https://resolver.caltech.edu/CaltechTHESIS:08312011-003358925
Year: 2012
DOI: 10.7907/MPHD-PC75
<p>Some of the most appealing science targets for future exploration missions in our solar system lie in terrains that are inaccessible to state-of-the-art robotic rovers such as NASA's Opportunity, thereby precluding in situ analysis of these rich opportunities. Examples of potential high-yield science areas on Mars include young gullies on sloped terrains, exposed layers of bedrock in the Victoria Crater, sources of methane gas near Martian volcanic ranges, and stepped delta formations in heavily cratered regions. In addition, a recently discovered cryovolcano on Titan and frozen water near the south pole of our own Moon could provide a wealth of knowledge to any robotic explorer capable of accessing these regions.</p>
<p>To address the challenge of extreme terrain exploration, this dissertation presents the Axel rover, a two-wheeled tethered robot capable of rappelling down steep slopes and traversing rocky terrain. Axel is part of a family of reconfigurable rovers, which, when docked, form a four-wheeled vehicle nicknamed DuAxel. DuAxel provides untethered mobility to regions of extreme terrain and serves as an anchor support for a single Axel when it undocks and rappels into low-ground.</p>
<p>Axel's performance on extreme terrain is primarily governed by three key system components: wheel design, tether control, and intelligent planning around obstacles. Investigations in wheel design and optimizing for extreme terrain resulted in the development of grouser wheels. Experiments demonstrated that these grouser wheels were very effective at surmounting obstacles, climbing rocks up to 90% of the wheel diameter. Terramechanics models supported by experiments showed that these wheels would not sink excessively or become trapped in deformable terrain.</p>
<p>Predicting tether forces in different configurations is also essential to the rover's mobility. Providing power, communication, and mobility forces, the tether is Axel's lifeline while it rappels steep slopes, and a cut, abraded, or ruptured tether would result in an untimely end to the rover's mission. Understanding tether forces are therefore paramount, and this thesis both models and measures tension forces to predict and avoid high-stress scenarios.</p>
<p>Finally, incorporating autonomy into Axel is a unique challenge due to the complications that arise during tether management. Without intelligent planning, rappelling systems can easily become entangled around obstacles and suffer catastrophic failures. This motivates the development of a novel tethered planning algorithm, presented in this thesis, which is unique for rappelling systems.</p>
<p>Recent field experiments in natural extreme terrains on Earth demonstrate the Axel rover's potential as a candidate for future space operations. Both DuAxel and its rappelling counterpart are rigorously tested on a 20 meter escarpment and in the Arizona desert. Through analysis and experiments, this thesis provides the framework for a new generation of robotic explorers capable of accessing extreme planetary regions and potentially providing clues for life beyond Earth.</p>https://resolver.caltech.edu/CaltechTHESIS:08312011-003358925MEMS for Glaucoma
https://resolver.caltech.edu/CaltechTHESIS:05192012-003206031
Year: 2012
DOI: 10.7907/4XD2-SP34
<p>Glaucoma is an eye disease that gradually steals vision. Open angle glaucoma is one of the most common glaucoma forms, in which eye fluid (aqueous humor) produced by the ciliary body cannot be drained away normally by patients’ eyes. The accumulated eye fluid inside the anterior chamber causes high intraocular pressure (IOP), which is transmitted onto the retina in the back of the eyeball (globe), continuously suppressing and damaging the patient’s optic nerves; this may lead to total blindness if not treated properly.</p>
<p>The current most-popular IOP monitoring technique is to use applanation tonometry, which applies applanation force onto the cornea and measures the resulting deformation in order to calculate the IOP. Even though applanation tonometry can provide quite useful information about patients’ IOP, continuous monitoring of IOP is required for ophthalmologists to understand the IOP fluctuation of the patients, something which still cannot be achieved via current applanation approach. In addition, applanation tonometry requires skillful operation performed by well-trained professionals, such as ophthalmologists, making continuous IOP monitoring impractical. In this work, we have developed a telemetric IOP sensor that is capable of monitoring IOP wirelessly and continuously. As the quality factor drops when a telemetric IOP sensor is implanted in the anterior chamber, due to the high loss tangent of the saline-based aqueous humor (~ 0.2) compared to air (0.0), a modified IOP sensor is developed to monitor IOP with sensing coil that is left exposed after implantation in order to avoid interruption from the eye fluid. Another approach is also proposed and tested to demonstrate that the quality factor can also be recovered by covering the sensing coil with low loss tangent materials.</p>
<p>Currently glaucoma is treated mostly by taking oral medications or applying eye drops. However, some glaucoma patients do not respond to those medications. Therefore, another physical approach, using a glaucoma drainage device (GDD), is necessary in order to drain out excessive eye fluid and serve as a long-term way to manage the increased IOP. Current commercially available glaucoma drainage devices do not have reliable valve systems to stop the drainage when the IOP falls into the normal range. Therefore, we have developed a dual-valved GDD to fulfill the “band-pass” flow regulation which drains out eye fluid only when IOP is higher than 20 mmHg, and stops drainage (closes the valve) when IOP is lower than 20 mmHg to prevent hypotony. The key component of GDD is a normally closed (NC) check-valve, which only opens to drain away the excess fluid when the pressure is higher than 20 mmHg. The proposed paradigm of our NC check-valve is to have a couple of parylene-C pre-stressed slanted tethers to provide the desired cracking pressure. The slanted tethers are achieved in this thesis by: 1) slanted photoresist generated by gray-scale photolithography, 2) pop-up mechanism, and 3) self-stiction bonding mechanism. The built-in residual tensile stress can be controlled by mechanical stretching or thermal annealing. The protecting mechanism preventing the unwanted drainage when the eyes experience sudden unpredicted high IOP is achieved by utilizing a normally open (NO) check-valve. A "minimally invasive implantation" procedure is proposed in the thesis to implant the GDD subconjunctivally. The small size of the device allows its insertion using a #19-gauge needle.</p>
<p>To accurately design the desired cracking pressure and also predict the lifetime of the NC check-valve, parylene-C’s mechanical, thermal, and polymer properties are investigated. The results show that the properties of parylene-C are highly process-temperature-dependent and therefore can be tailored by adjusting the thermal annealing process.</p>https://resolver.caltech.edu/CaltechTHESIS:05192012-003206031Parylene as a New Membrane Material for Biomems Applications
https://resolver.caltech.edu/CaltechTHESIS:05022012-003225468
Year: 2012
DOI: 10.7907/DPDC-9E57
<p>The work in this thesis aims to use MEMS and microfabrication technologies to develop two types of parylene membrane devices for biomedical applications. The first device is the parylene membrane filter for cancer detection. The presence of circulating tumor cells (CTC) in patient blood is an important sign of cancer metastasis. However, currently there are two big challenges for CTC detection. First, CTCs are extremely rare, especially at the early stage of cancer metastasis. Secondly, CTCs are very fragile, and are very likely to be damaged during the capturing process. By using size-based membrane filtration through the specially designed parylene filters, together with a constant-pressure filtration system, we are able to capture the CTCs from patient blood with high capture efficiency, high viability, moderate enrichment, and high throughput. Both immunofluorescence enumeration and telomerase activity detection have been used to detect and differentiate the captured CTCs. The feasibility of further cell culture of the captured CTCs has also been demonstrated, which could be a useful way to increase the number of CTCs for future studies. Models of the time-dependent cell membrane damage are developed to predict and prevent CTC damage during this detection process. The results of clinical trials further demonstrate that the parylene membrane filter is a promising device for cancer detection.</p>
<p>The second device is the parylene artificial Bruch’s membrane for age-related macular degeneration (AMD). AMD is usually characterized by an impaired Bruch’s membrane with much lowered permeability, which impedes the transportation of nutrients from choroid vessels to nourish the retinal pigment epithelial (RPE) cells and photoreceptors. Parylene is selected as a substitute material because of its good mechanical properties, transparency, biocompatibility, and machinability. More importantly, it is found that the permeability of submicron parylene is very similar to that of healthy human Bruch’s membrane. A mesh-supported submicron parylene membrane structure has been designed and its feasibility as an artificial Bruch’s membrane has been demonstrated by diffusion experiments, cell perfusion culture, and pressure deflection tests. RPE cells are able to adhere, proliferate and develop into normal in vivo-like morphology and functions. Currently this artificial membrane is under clinical trials.</p>
https://resolver.caltech.edu/CaltechTHESIS:05022012-003225468Flight Dynamics in Drosophila Through a Dynamically-scaled Robotic Approach
https://resolver.caltech.edu/CaltechTHESIS:06072013-110839676
Year: 2013
DOI: 10.7907/MSRS-JG88
<p>Flies are particularly adept at balancing the competing demands of delay tolerance, performance, and robustness during flight, which invites thoughtful examination of their multimodal feedback architecture. This dissertation examines stabilization requirements for inner-loop feedback strategies in the flapping flight of Drosophila, the fruit fly, against the backdrop of sensorimotor transformations present in the animal. Flies have evolved multiple specializations to reduce sensorimotor latency, but sensory delay during flight is still significant on the timescale of body dynamics. I explored the effect of sensor delay on flight stability and performance for yaw turns using a dynamically-scaled robot equipped with a real-time feedback system that performed active turns in response to measured yaw torque. The results show a fundamental tradeoff between sensor delay and permissible feedback gain, and suggest that fast mechanosensory feedback provides a source of active damping that compliments that contributed by passive effects. Presented in the context of these findings, a control architecture whereby a haltere-mediated inner-loop proportional controller provides damping for slower visually-mediated feedback is consistent with tethered-flight measurements, free-flight observations, and engineering design principles.</p>
<p>Additionally, I investigated how flies adjust stroke features to regulate and stabilize level forward flight. The results suggest that few changes to hovering kinematics are actually required to meet steady-state lift and thrust requirements at different flight speeds, and the primary driver of equilibrium velocity is the aerodynamic pitch moment. This finding is consistent with prior hypotheses and observations regarding the relationship between body pitch and flight speed in fruit flies. The results also show that the dynamics may be stabilized with additional pitch damping, but the magnitude of required damping increases with flight speed. I posit that differences in stroke deviation between the upstroke and downstroke might play a critical role in this stabilization. Fast mechanosensory feedback of the pitch rate could enable active damping, which would inherently exhibit gain scheduling with flight speed if pitch torque is regulated by adjusting stroke deviation. Such a control scheme would provide an elegant solution for flight stabilization across a wide range of flight speeds.</p>
https://resolver.caltech.edu/CaltechTHESIS:06072013-110839676Parylene-C as a New Piezoelectric Material
https://resolver.caltech.edu/CaltechTHESIS:05142013-075702623
Year: 2013
DOI: 10.7907/1VEH-EP90
<p>The goal of this thesis is to develop a proper microelectromechanical systems (MEMS) process to manufacture piezoelectric Parylene-C (PA-C), which is famous for its chemical inertness, mechanical and thermal properties and electrical insulation. Furthermore, piezoelectric PA-C is used to build miniature, inexpensive, non-biased piezoelectric microphones.</p>
<p>These piezoelectric PA-C MEMS microphones are to be used in any application where a conventional piezoelectric and electret microphone can be used, such as in cell phones and hearing aids. However, they have the advantage of a simplified fabrication process compared with existing technology. In addition, as a piezoelectric polymer, PA-C has varieties of applications due to its low dielectric constant, low elastic stiffness, low density, high voltage sensitivity, high temperature stability and low acoustic and mechanical impedance. Furthermore, PA-C is an FDA approved biocompatible material and is able to maintain operate at a high temperature.</p>
<p>To accomplish piezoelectric PA-C, a MEMS-compatible poling technology has been developed. The PA-C film is poled by applying electrical field during heating. The piezoelectric coefficient, -3.75pC/N, is obtained without film stretching.</p>
<p>The millimeter-scale piezoelectric PA-C microphone is fabricated with an in-plane spiral arrangement of two electrodes. The dynamic range is from less than 30 dB to above 110 dB SPL (referenced 20 µPa) and the open-circuit sensitivities are from 0.001 – 0.11 mV/Pa over a frequency range of 1 - 10 kHz. The total harmonic distortion of the device is less than 20% at 110 dB SPL and 1 kHz.</p>https://resolver.caltech.edu/CaltechTHESIS:05142013-075702623Design, Specification, and Synthesis of Aircraft Electric Power Systems Control Logic
https://resolver.caltech.edu/CaltechTHESIS:05312013-103940337
Year: 2013
DOI: 10.7907/QDJN-BB72
<p>Cyber-physical systems integrate computation, networking, and physical processes. Substantial research challenges exist in the design and verification of such large-scale, distributed sensing, ac- tuation, and control systems. Rapidly improving technology and recent advances in control theory, networked systems, and computer science give us the opportunity to drastically improve our approach to integrated flow of information and cooperative behavior. Current systems rely on text-based spec- ifications and manual design. Using new technology advances, we can create easier, more efficient, and cheaper ways of developing these control systems. This thesis will focus on design considera- tions for system topologies, ways to formally and automatically specify requirements, and methods to synthesize reactive control protocols, all within the context of an aircraft electric power system as a representative application area.</p>
<p>This thesis consists of three complementary parts: synthesis, specification, and design. The first section focuses on the synthesis of central and distributed reactive controllers for an aircraft elec- tric power system. This approach incorporates methodologies from computer science and control. The resulting controllers are correct by construction with respect to system requirements, which are formulated using the specification language of linear temporal logic (LTL). The second section addresses how to formally specify requirements and introduces a domain-specific language for electric power systems. A software tool automatically converts high-level requirements into LTL and synthesizes a controller.</p>
<p>The final sections focus on design space exploration. A design methodology is proposed that uses mixed-integer linear programming to obtain candidate topologies, which are then used to synthesize controllers. The discrete-time control logic is then verified in real-time by two methods: hardware and simulation. Finally, the problem of partial observability and dynamic state estimation is ex- plored. Given a set placement of sensors on an electric power system, measurements from these sensors can be used in conjunction with control logic to infer the state of the system.</p>https://resolver.caltech.edu/CaltechTHESIS:05312013-103940337Estimation and Inference for Grasping and Manipulation Tasks Using Vision and Kinesthetic Sensors
https://resolver.caltech.edu/CaltechTHESIS:04052013-105520483
Year: 2013
DOI: 10.7907/PZB6-QJ39
<p>This thesis presents a novel framework for state estimation in the context of robotic grasping and manipulation. The overall estimation approach is based on fusing various visual cues for manipulator tracking, namely appearance and feature-based, shape-based, and silhouette-based visual cues. Similarly, a framework is developed to fuse the above visual cues, but also kinesthetic cues such as force-torque and tactile measurements, for in-hand object pose estimation. The cues are extracted from multiple sensor modalities and are fused in a variety of Kalman filters.</p>
<p>A hybrid estimator is developed to estimate both a continuous state (robot and object states) and discrete states, called contact modes, which specify how each finger contacts a particular object surface. A static multiple model estimator is used to compute and maintain this mode probability. The thesis also develops an estimation framework for estimating model parameters associated with object grasping. Dual and joint state-parameter estimation is explored for parameter estimation of a grasped object's mass and center of mass. Experimental results demonstrate simultaneous object localization and center of mass estimation.</p>
<p>Dual-arm estimation is developed for two arm robotic manipulation tasks. Two types of filters are explored; the first is an augmented filter that contains both arms in the state vector while the second runs two filters in parallel, one for each arm. These two frameworks and their performance is compared in a dual-arm task of removing a wheel from a hub.</p>
<p>This thesis also presents a new method for action selection involving touch. This next best touch method selects an available action for interacting with an object that will gain the most information. The algorithm employs information theory to compute an information gain metric that is based on a probabilistic belief suitable for the task. An estimation framework is used to maintain this belief over time. Kinesthetic measurements such as contact and tactile measurements are used to update the state belief after every interactive action. Simulation and experimental results are demonstrated using next best touch for object localization, specifically a door handle on a door.
The next best touch theory is extended for model parameter determination. Since many objects within a particular object category share the same rough shape, principle component analysis may be used to parametrize the object mesh models. These parameters can be estimated using the action selection technique that selects the touching action which best both localizes and estimates these parameters. Simulation results are then presented involving localizing and determining a parameter of a screwdriver.</p>
<p>Lastly, the next best touch theory is further extended to model classes. Instead of estimating parameters, object class determination is incorporated into the information gain metric calculation. The best touching action is selected in order to best discern between the possible model classes. Simulation results are presented to validate the theory.</p>https://resolver.caltech.edu/CaltechTHESIS:04052013-105520483Robot Navigation in Dense Crowds: Statistical Models and Experimental Studies of Human Robot Cooperation
https://resolver.caltech.edu/CaltechTHESIS:05182013-191132413
Year: 2013
DOI: 10.7907/BHGM-0C65
<p>This thesis explores the problem of mobile robot navigation in dense human crowds. We begin by considering a fundamental impediment to classical motion planning algorithms called the freezing robot problem: once the environment surpasses a certain level of complexity, the planner decides that all forward paths are unsafe, and the robot freezes in place (or performs unnecessary maneuvers) to avoid collisions. Since a feasible path typically exists, this behavior is suboptimal. Existing approaches have focused on reducing predictive uncertainty by employing higher fidelity individual dynamics models or heuristically limiting the individual predictive covariance to prevent overcautious navigation. We demonstrate that both the individual prediction and the individual predictive uncertainty have little to do with this undesirable navigation behavior. Additionally, we provide evidence that dynamic agents are able to navigate in dense crowds by engaging in joint collision avoidance, cooperatively making room to create feasible trajectories. We accordingly develop interacting Gaussian processes, a prediction density that captures cooperative collision avoidance, and a "multiple goal" extension that models the goal driven nature of human decision making. Navigation naturally emerges as a statistic of this distribution.</p>
<p>Most importantly, we empirically validate our models in the Chandler dining hall at Caltech during peak hours, and in the process, carry out the first extensive quantitative study of robot navigation in dense human crowds (collecting data on 488 runs). The multiple goal interacting Gaussian processes algorithm performs comparably with human teleoperators in crowd densities nearing 1 person/m<sup>2</sup>, while a state of the art noncooperative planner exhibits unsafe behavior more than 3 times as often as the multiple goal extension, and twice as often as the basic interacting Gaussian process approach. Furthermore, a reactive planner based on the widely used dynamic window approach proves insufficient for crowd densities above 0.55 people/m<sup>2</sup>. We also show that our noncooperative planner or our reactive planner capture the salient characteristics of nearly any dynamic navigation algorithm. For inclusive validation purposes, we show that either our non-interacting planner or our reactive planner captures the salient characteristics of nearly any existing dynamic navigation algorithm. Based on these experimental results and theoretical observations, we conclude that a cooperation model is critical for safe and efficient robot navigation in dense human crowds.</p>
<p>Finally, we produce a large database of ground truth pedestrian crowd data. We make this ground truth database publicly available for further scientific study of crowd prediction models, learning from demonstration algorithms, and human robot interaction models in general.</p>
https://resolver.caltech.edu/CaltechTHESIS:05182013-191132413Blood Cell Count On-a-Chip
https://resolver.caltech.edu/CaltechTHESIS:06252012-171451630
Year: 2013
DOI: 10.7907/6YF1-WR04
<p>White blood cell (WBC) count is one of the most frequently ordered clinical tests in hospitals. There are five types of WBCs in the circulating blood, including lymphocyte, monocyte, neutrophil, eosinophil and basophil. The WBC count test enumerates not only the total number of WBCs in per volume blood, but also the percentage of each WBC type. A portable instrument for the WBC count test is currently in demand by the NASA human spaceflight, and also by the on-earth telemedicine application. However, the commercially available tests do not meet the requirement of the portable applications, because of their large instrument size and the large reagent volume consumed per test. </p>
<p>This study describes the development of a WBC count technology optimized for portable applications. First, a sheathless microfluidic cytometer is developed for WBC count. This technology consumes only a small amount of blood (5 microlitre) and a minimal volume of reagents (50 microlitre). Second, fluorescent dye assays are developed for the WBC differential count by measuring fluorescent emissions on the microfluidic cytometer. Based on this technology, a portable instrument is built with high test accuracy (maximum error less than 10%).</p>
<p>Furthermore, this study explores two key components for future integrating this technology into a self-contained chip. First, a microvalve actuated by thermal blood clogging is developed. This valve has a simple structure suitable for on-chip integration. Second, a micromixer is used to demonstrate the staining of blood with dye assays, and the following fluorescent detection of WBCs on the cytometer.</p>
https://resolver.caltech.edu/CaltechTHESIS:06252012-171451630Bootstrapping Vehicles: A Formal Approach to Unsupervised Sensorimotor Learning Based on Invariance
https://resolver.caltech.edu/CaltechTHESIS:10282012-082208075
Year: 2013
DOI: 10.7907/PWVS-2Q74
Could a "brain in a jar" be able to control an unknown robotic body to which it is connected, and use it to achieve useful tasks, without any prior assumptions on the body's sensors and actuators? Other than of purely intellectual interest, this question is relevant to the medium-term challenges of robotics: as the complexity of robotics applications grows, automated learning techniques might reduce design effort and increase the robustness and reliability of the solutions. In this work, the problem of "bootstrapping" is studied in the context of the Vehicles universe, which is an idealization of simple mobile robots, after the work of Braitenberg. The first thread of results consists in analyzing such simple sensorimotor cascades and proposing models of varying complexity that can be learned from data. The second thread regards how to properly formalize the notions of "absence of assumptions", as a particular form of invariance that the bootstrapping agent must satisfy, and proposes some invariance-based design techniques.https://resolver.caltech.edu/CaltechTHESIS:10282012-082208075Spinal Cord Injury Therapy through Active Learning
https://resolver.caltech.edu/CaltechTHESIS:07252013-120308708
Year: 2014
DOI: 10.7907/X5M7-EC09
Therapy employing epidural electrostimulation holds great potential for improving therapy for patients with spinal cord injury (SCI) (Harkema et al., 2011). Further promising results from combined therapies using electrostimulation have also been recently obtained (e.g., van den Brand et al., 2012). The devices being developed to deliver the stimulation are highly flexible, capable of delivering any individual stimulus among a combinatorially large set of stimuli (Gad et al., 2013). While this extreme flexibility is very useful for ensuring that the device can deliver an appropriate stimulus, the challenge of choosing good stimuli is quite substantial, even for expert human experimenters. To develop a fully implantable, autonomous device which can provide useful therapy, it is necessary to design an algorithmic method for choosing the stimulus parameters. Such a method can be used in a clinical setting, by caregivers who are not experts in the neurostimulator's use, and to allow the system to adapt autonomously between visits to the clinic. To create such an algorithm, this dissertation pursues the general class of active learning algorithms that includes Gaussian Process Upper Confidence Bound (GP-UCB, Srinivas et al., 2010), developing the Gaussian Process Batch Upper Confidence Bound (GP-BUCB, Desautels et al., 2012) and Gaussian Process Adaptive Upper Confidence Bound (GP-AUCB) algorithms. This dissertation develops new theoretical bounds for the performance of these and similar algorithms, empirically assesses these algorithms against a number of competitors in simulation, and applies a variant of the GP-BUCB algorithm in closed-loop to control SCI therapy via epidural electrostimulation in four live rats. The algorithm was tasked with maximizing the amplitude of evoked potentials in the rats' left tibialis anterior muscle. These experiments show that the algorithm is capable of directing these experiments sensibly, finding effective stimuli in all four animals. Further, in direct competition with an expert human experimenter, the algorithm produced superior performance in terms of average reward and comparable or superior performance in terms of maximum reward. These results indicate that variants of GP-BUCB may be suitable for autonomously directing SCI therapy.https://resolver.caltech.edu/CaltechTHESIS:07252013-120308708Wireless Parylene-Based Retinal Implant
https://resolver.caltech.edu/CaltechTHESIS:08202013-193131878
Year: 2014
DOI: 10.7907/YTN7-ZA05
<p>The degeneration of the outer retina usually causes blindness by affecting the photoreceptor cells. However, the ganglion cells, which consist of optic nerves, on the middle and inner retina layers are often intact. The retinal implant, which can partially restore vision by electrical stimulation, soon becomes a focus for research. Although many groups worldwide have spent a lot of effort on building devices for retinal implant, current state-of-the-art technologies still lack a reliable packaging scheme for devices with desirable high-density multi-channel features. Wireless flexible retinal implants have always been the ultimate goal for retinal prosthesis. In this dissertation, the reliable packaging scheme for a wireless flexible parylene-based retinal implants has been well developed. It can not only provide stable electrical and mechanical connections to the high-density multi-channel (1000+ channels on 5 mm × 5 mm chip area) IC chips, but also survive for more than 10 years in the human body with corrosive fluids.</p>
<p>The device is based on a parylene-metal-parylene sandwich structure. In which, the adhesion between the parylene layers and the metals embedded in the parylene layers have been studied. Integration technology for high-density multi-channel IC chips has also been addressed and tested with dummy and real 268-channel and 1024-channel retinal IC chips. In addition, different protection schemes have been tried in application to IC chips and discrete components to gain the longest lifetime. The effectiveness has been confirmed by the accelerated and active lifetime soaking test in saline solution. Surgical mockups have also been designed and successfully implanted inside dog's and pig's eyes. Additionally, the electrodes used to stimulate the ganglion cells have been modified to lower the interface impedance and shaped to better fit the retina. Finally, all the developed technologies have been applied on the final device with a dual-metal-layer structure.</p>
https://resolver.caltech.edu/CaltechTHESIS:08202013-193131878Formal Methods for Control Synthesis in Partially Observed Environments: Application to Autonomous Robotic Manipulation
https://resolver.caltech.edu/CaltechTHESIS:05292014-063852576
Year: 2014
DOI: 10.7907/RQKC-N871
<p>Modern robots are increasingly expected to function in uncertain and dynamically challenging environments, often in proximity with humans. In addition, wide scale adoption of robots requires on-the-fly adaptability of software for diverse application. These requirements strongly suggest the need to adopt formal representations of high level goals and safety specifications, especially as temporal logic formulas. This approach allows for the use of formal verification techniques for controller synthesis that can give guarantees for safety and performance. Robots operating in unstructured environments also face limited sensing capability. Correctly inferring a robot's progress toward high level goal can be challenging.</p>
<p>This thesis develops new algorithms for synthesizing discrete controllers in partially known environments under specifications represented as linear temporal logic (LTL) formulas. It is inspired by recent developments in finite abstraction techniques for hybrid systems and motion planning problems. The robot and its environment is assumed to have a finite abstraction as a Partially Observable Markov Decision Process (POMDP), which is a powerful model class capable of representing a wide variety of problems. However, synthesizing controllers that satisfy LTL goals over POMDPs is a challenging problem which has received only limited attention.</p>
<p>This thesis proposes tractable, approximate algorithms for the control synthesis problem using Finite State Controllers (FSCs). The use of FSCs to control finite POMDPs allows for the closed system to be analyzed as finite global Markov chain. The thesis explicitly shows how transient and steady state behavior of the global Markov chains can be related to two different criteria with respect to satisfaction of LTL formulas. First, the maximization of the probability of LTL satisfaction is related to an optimization problem over a parametrization of the FSC. Analytic computation of gradients are derived which allows the use of first order optimization techniques. </p>
<p>The second criterion encourages rapid and frequent visits to a restricted set of states over infinite executions. It is formulated as a constrained optimization problem with a discounted long term reward objective by the novel utilization of a fundamental equation for Markov chains - the Poisson equation. A new constrained policy iteration technique is proposed to solve the resulting dynamic program, which also provides a way to escape local maxima.</p>
<p>The algorithms proposed in the thesis are applied to the task planning and execution challenges faced during the DARPA Autonomous Robotic Manipulation - Software challenge.</p>https://resolver.caltech.edu/CaltechTHESIS:05292014-063852576Complex Behavior and Perception in Drosophila Emerges from Iterative Feedback-Regulated Reflexes
https://resolver.caltech.edu/CaltechTHESIS:01032014-164946523
Year: 2014
DOI: 10.7907/WSE4-WG98
<p>For a hungry fruit fly, locating and landing on a fermenting fruit where it can feed, find mates, and lay eggs, is an essential and difficult task requiring the integration of both olfactory and visual cues. Understanding how flies accomplish this will help provide a comprehensive ethological context for the expanding knowledge of their neural circuits involved in processing olfaction and vision, as well as inspire novel engineering solutions for control and estimation in computationally limited robotic applications. In this thesis, I use novel high throughput methods to develop a detailed overview of how flies track odor plumes, land, and regulate flight speed. Finally, I provide an example of how these insights can be applied to robotic applications to simplify complicated estimation problems. To localize an odor source, flies exhibit three iterative, reflex-driven behaviors. Upon encountering an attractive plume, flies increase their flight speed and turn upwind using visual cues. After losing the plume, flies begin zigzagging crosswind, again using visual cues to control their heading. After sensing an attractive odor, flies become more attracted to small visual features, which increases their chances of finding the plume source. Their changes in heading are largely controlled by open-loop maneuvers called saccades, which they direct towards and away from visual features. If a fly decides to land on an object, it begins to decelerate so as to maintain a stereotypical ratio of expansion to retinal size. Once they reach a stereotypical distance from the target, flies extend their legs in preparation for touchdown. Although it is unclear what cues they use to trigger this behavior, previous studies have indicated that it is likely under visual control. In Chapter 3, I use a nonlinear control theoretic analysis and robotic testbed to propose a novel and putative mechanism for how a fly might visually estimate distance by actively decelerating according to a visual control law. Throughout these behaviors, a common theme is the visual control of flight speed. Using genetic tools I show that the neuromodulator octopamine plays an important role in regulating flight speed, and propose a neural circuit for how this controller might be implemented in the flies brain. Two general biological and engineering principles are evident across my experiments: (1) complex behaviors, such as foraging, can emerge from the interactions of simple independent sensory-motor modules; (2) flies control their behavior in such a way that simplifies complex estimation problems.</p>https://resolver.caltech.edu/CaltechTHESIS:01032014-164946523Control of Dynamical Systems with Temporal Logic Specifications
https://resolver.caltech.edu/CaltechTHESIS:02172014-121159358
Year: 2014
DOI: 10.7907/TGFR-SS39
<p>This thesis is motivated by safety-critical applications involving autonomous air, ground, and space vehicles carrying out complex tasks in uncertain and adversarial environments. We use temporal logic as a language to formally specify complex tasks and system properties. Temporal logic specifications generalize the classical notions of stability and reachability that are studied in the control and hybrid systems communities. Given a system model and a formal task specification, the goal is to automatically synthesize a control policy for the system that ensures that the system satisfies the specification. This thesis presents novel control policy synthesis algorithms for optimal and robust control of dynamical systems with temporal logic specifications. Furthermore, it introduces algorithms that are efficient and extend to high-dimensional dynamical systems.</p>
<p>The first contribution of this thesis is the generalization of a classical linear temporal logic (LTL) control synthesis approach to optimal and robust control. We show how we can extend automata-based synthesis techniques for discrete abstractions of dynamical systems to create optimal and robust controllers that are guaranteed to satisfy an LTL specification. Such optimal and robust controllers can be computed at little extra computational cost compared to computing a feasible controller.</p>
<p>The second contribution of this thesis addresses the scalability of control synthesis with LTL specifications. A major limitation of the standard automaton-based approach for control with LTL specifications is that the automaton might be doubly-exponential in the size of the LTL specification. We introduce a fragment of LTL for which one can compute feasible control policies in time polynomial in the size of the system and specification. Additionally, we show how to compute optimal control policies for a variety of cost functions, and identify interesting cases when this can be done in polynomial time. These techniques are particularly relevant for online control, as one can guarantee that a feasible solution can be found quickly, and then iteratively improve on the quality as time permits.</p>
<p>The final contribution of this thesis is a set of algorithms for computing feasible trajectories for high-dimensional, nonlinear systems with LTL specifications. These algorithms avoid a potentially computationally-expensive process of computing a discrete abstraction, and instead compute directly on the system's continuous state space. The first method uses an automaton representing the specification to directly encode a series of constrained-reachability subproblems, which can be solved in a modular fashion by using standard techniques. The second method encodes an LTL formula as mixed-integer linear programming constraints on the dynamical system. We demonstrate these approaches with numerical experiments on temporal logic motion planning problems with high-dimensional (10+ states) continuous systems.</p>https://resolver.caltech.edu/CaltechTHESIS:02172014-121159358Efficient Methods for Stochastic Optimal Control
https://resolver.caltech.edu/CaltechTHESIS:05312014-011052261
Year: 2014
DOI: 10.7907/D40A-9E03
<p>The Hamilton Jacobi Bellman (HJB) equation is central to stochastic optimal control (SOC) theory, yielding the optimal solution to general problems specified by known dynamics and a specified cost functional. Given the assumption of quadratic cost on the control input, it is well known that the HJB reduces to a particular partial differential equation (PDE). While powerful, this reduction is not commonly used as the PDE is of second order, is nonlinear, and examples exist where the problem may not have a solution in a classical sense. Furthermore, each state of the system appears as another dimension of the PDE, giving rise to the curse of dimensionality. Since the number of degrees of freedom required to solve the optimal control problem grows exponentially with dimension, the problem becomes intractable for systems with all but modest dimension.</p>
<p>In the last decade researchers have found that under certain, fairly non-restrictive structural assumptions, the HJB may be transformed into a linear PDE, with an interesting analogue in the discretized domain of Markov Decision Processes (MDP). The work presented in this thesis uses the linearity of this particular form of the HJB PDE to push the computational boundaries of stochastic optimal control.</p>
<p>This is done by crafting together previously disjoint lines of research in computation. The first of these is the use of Sum of Squares (SOS) techniques for synthesis of control policies. A candidate polynomial with variable coefficients is proposed as the solution to the stochastic optimal control problem. An SOS relaxation is then taken to the partial differential constraints, leading to a hierarchy of semidefinite relaxations with improving sub-optimality gap. The resulting approximate solutions are shown to be guaranteed over- and under-approximations for the optimal value function. It is shown that these results extend to arbitrary parabolic and elliptic PDEs, yielding a novel method for Uncertainty Quantification (UQ) of systems governed by partial differential constraints. Domain decomposition techniques are also made available, allowing for such problems to be solved via parallelization and low-order polynomials.</p>
<p>The optimization-based SOS technique is then contrasted with the Separated Representation (SR) approach from the applied mathematics community. The technique allows for systems of equations to be solved through a low-rank decomposition that results in algorithms that scale linearly with dimensionality. Its application in stochastic optimal control allows for previously uncomputable problems to be solved quickly, scaling to such complex systems as the Quadcopter and VTOL aircraft. This technique may be combined with the SOS approach, yielding not only a numerical technique, but also an analytical one that allows for entirely new classes of systems to be studied and for stability properties to be guaranteed.</p>
<p>The analysis of the linear HJB is completed by the study of its implications in application. It is shown that the HJB and a popular technique in robotics, the use of navigation functions, sit on opposite ends of a spectrum of optimization problems, upon which tradeoffs may be made in problem complexity. Analytical solutions to the HJB in these settings are available in simplified domains, yielding guidance towards optimality for approximation schemes. Finally, the use of HJB equations in temporal multi-task planning problems is investigated. It is demonstrated that such problems are reducible to a sequence of SOC problems linked via boundary conditions. The linearity of the PDE allows us to pre-compute control policy primitives and then compose them, at essentially zero cost, to satisfy a complex temporal logic specification.</p> https://resolver.caltech.edu/CaltechTHESIS:05312014-011052261Multilayer Active Shell Mirrors
https://resolver.caltech.edu/CaltechTHESIS:05282015-145339959
Year: 2015
DOI: 10.7907/Z99W0CFB
<p>This thesis presents a novel active mirror technology based on carbon fiber composites and replication manufacturing processes. Multiple additional layers are implemented into the structure in order to provide the reflective layer, actuation capabilities and electrode routing. The mirror is thin, lightweight, and has large actuation capabilities. These features, along with the associated manufacturing processes, represent a significant change in design compared to traditional optics. Structural redundancy in the form of added material or support structures is replaced by thin, unsupported lightweight substrates with large actuation capabilities.</p>
<p>Several studies motivated by the desire to improve as-manufactured figure quality are performed. Firstly, imperfections in thin CFRP laminates and their effect on post-cure shape errors are studied. Numerical models are developed and compared to experimental measurements on flat laminates. Techniques to mitigate figure errors for thicker laminates are also identified. A method of properly integrating the reflective facesheet onto the front surface of the CFRP substrate is also presented. Finally, the effect of bonding multiple initially flat active plates to the backside of a curved CFRP substrate is studied. Figure deformations along with local surface defects are predicted and characterized experimentally. By understanding the mechanics behind these processes, significant improvements to the overall figure quality have been made. </p>
<p>Studies related to the actuation response of the mirror are also performed. The active properties of two materials are characterized and compared. Optimal active layer thicknesses for thin surface-parallel schemes are determined. Finite element simulations are used to make predictions on shape correction capabilities, demonstrating high correctabiliity and stroke over low-order modes. The effect of actuator saturation is studied and shown to significantly degrade shape correction performance.</p>
<p>The initial figure as well as actuation capabilities of a fully-integrated active mirror prototype are characterized experimentally using a Projected Hartmann test. A description of the test apparatus is presented along with two verification measurements. The apparatus is shown to accurately capture both high-amplitude low spatial-frequency figure errors as well as those at lower amplitudes but higher spatial frequencies. A closed-loop figure correction is performed, reducing figure errors by 94%.</p>https://resolver.caltech.edu/CaltechTHESIS:05282015-145339959On the Role of Delays in Biological Systems : Analysis and Design
https://resolver.caltech.edu/CaltechTHESIS:08282014-165029252
Year: 2015
DOI: 10.7907/Z9JH3J4W
<p>This work quantifies the nature of delays in genetic regulatory networks and their effect on system dynamics. It is known that a time lag can emerge from a sequence of biochemical reactions. Applying this modeling framework to the protein production processes, delay distributions are derived in a stochastic (probability density function) and deterministic setting (impulse function), whilst being shown to be equivalent under different assumptions. The dependence of the distribution properties on rate constants, gene length, and time-varying temperatures is investigated. Overall, the distribution of the delay in the context of protein production processes is shown to be highly dependent on the size of the genes and mRNA strands as well as the reaction rates. Results suggest longer genes have delay distributions with a smaller relative variance, and hence, less uncertainty in the completion times, however, they lead to larger delays. On the other hand large uncertainties may actually play a positive role, as broader distributions can lead to larger stability regions when this formalization of the protein production delays is incorporated into a feedback system.</p>
<p>Furthermore, evidence suggests that delays may play a role as an explicit design into existing controlling mechanisms. Accordingly, the reccurring dual-feedback motif is also investigated with delays incorporated into the feedback channels. The dual-delayed feedback is shown to have stabilizing effects through a control theoretic approach. Lastly, a distributed delay based controller design method is proposed as a potential design tool. In a preliminary study, the dual-delayed feedback system re-emerges as an effective controller design.</p> https://resolver.caltech.edu/CaltechTHESIS:08282014-165029252Microelectrode Implants for Spinal Cord Stimulation in Rats
https://resolver.caltech.edu/CaltechTHESIS:06052015-084726649
Year: 2015
DOI: 10.7907/Z9930R3G
<p>Paralysis is a debilitating condition afflicting millions of people across the globe, and is particularly deleterious to quality of life when motor function of the legs is severely impaired or completely absent. Fortunately, spinal cord stimulation has shown great potential for improving motor function after spinal cord injury and other pathological conditions. Many animal studies have shown stimulation of the neural networks in the spinal cord can improve motor ability so dramatically that the animals can even stand and step after a complete spinal cord transaction.</p>
<p>This thesis presents work to successfully provide a chronically implantable device for rats that greatly enhances the ability to control the site of spinal cord stimulation. This is achieved through the use of a parylene-C based microelectrode array, which enables a density of stimulation sites unattainable with conventional wire electrodes. While many microelectrode devices have been proposed in the past, the spinal cord is a particularly challenging environment due to the bending and movement it undergoes in a live animal. The developed microelectrode array is the first to have been implanted in vivo while retaining functionality for over a month. In doing so, different neural pathways can be selectively activated to facilitate standing and stepping in spinalized rats using various electrode combinations, and important differences in responses are observed.</p>
<p>An engineering challenge for the usability of any high density electrode array is connecting the numerous electrodes to a stimulation source. This thesis develops several technologies to address this challenge, beginning with a fully passive implant that uses one wire per electrode to connect to an external stimulation source. The number of wires passing through the body and the skin proved to be a hazard for the health of the animal, so a multiplexed implant was devised in which active electronics reduce the number of wires. Finally, a fully wireless implant was developed. As these implants are tested in vivo, encapsulation is of critical importance to retain functionality in a chronic experiment, especially for the active implants, and it was achieved without the use of costly ceramic or metallic hermetic packaging. Active implants were built that retained functionality 8 weeks after implantation, and achieved stepping in spinalized rats after just 8-10 days, which is far sooner than wire-based electrical stimulation has achieved in prior work.</p>https://resolver.caltech.edu/CaltechTHESIS:06052015-084726649Kinematics and Local Motion Planning for Quasi-static Whole-body Mobile Manipulation
https://resolver.caltech.edu/CaltechTHESIS:05222016-095145651
Year: 2016
DOI: 10.7907/Z9KK98RX
<p>This thesis studies mobile robotic manipulators, where one or more robot manipulator arms are
integrated with a mobile robotic base. The base could be a wheeled or tracked vehicle, or it might be a
multi-limbed locomotor. As robots are increasingly deployed in complex and unstructured environments,
the need for mobile manipulation increases. Mobile robotic assistants have the potential to revolutionize human
lives in a large variety of settings including home, industrial and outdoor environments.</p>
<p>Mobile Manipulation is the use or study of such mobile robots as they interact with physical
objects in their environment. As compared to fixed base manipulators, mobile manipulators can take
advantage of the base mechanism’s added degrees of freedom in the task planning and execution process.
But their use also poses new problems in the analysis and control of base system stability, and the
planning of coordinated base and arm motions. For mobile manipulators to be successfully and
efficiently used, a thorough understanding of their kinematics, stability, and capabilities is required.
Moreover, because mobile manipulators typically possess a large number of actuators, new and efficient
methods to coordinate their large numbers of degrees of freedom are needed to make them practically
deployable. This thesis develops new kinematic and stability analyses of mobile manipulation, and new
algorithms to efficiently plan their motions.</p>
<p>I first develop detailed and novel descriptions of the kinematics governing the operation of multi-
limbed legged robots working in the presence of gravity, and whose limbs may also be simultaneously
used for manipulation. The fundamental stance constraint that arises from simple assumptions about
friction and the ground contact and feasible motions is derived. Thereafter, a local relationship between
joint motions and motions of the robot abdomen and reaching limbs is developed. Baseeon these
relationships, one can define and analyze local kinematic qualities including limberness, wrench
resistance and local dexterity. While previous researchers have noted the similarity between multi-
fingered grasping and quasi-static manipulation, this thesis makes explicit connections between these two
problems.</p>
<p>The kinematic expressions form the basis for a local motion planning problem that that
determines the joint motions to achieve several simultaneous objectives while maintaining stance stability
in the presence of gravity. This problem is translated into a convex quadratic program entitled the
balanced priority solution, whose existence and uniqueness properties are developed. This problem is
related in spirit to the classical redundancy resoxlution and task-priority approaches. With some simple
modifications, this local planning and optimization problem can be extended to handle a large variety of
goals and constraints that arise in mobile-manipulation. This local planning problem applies readily to
other mobile bases including wheeled and articulated bases. This thesis describes the use of the local
planning techniques to generate global plans, as well as for use within a feedback loop. The work in this
thesis is motivated in part by many practical tasks involving the Surrogate and RoboSimian robots at
NASA/JPL, and a large number of examples involving the two robots, both real and simulated, are
provided.</p>
<p>Finally, this thesis provides an analysis of simultaneous force and motion control for multi-
limbed legged robots. Starting with a classical linear stiffness relationship, an analysis of this problem for
multiple point contacts is described. The local velocity planning problem is extended to include
generation of forces, as well as to maintain stability using force-feedback. This thesis also provides a
concise, novel definition of static stability, and proves some conditions under which it is satisfied.</p>https://resolver.caltech.edu/CaltechTHESIS:05222016-095145651Incremental Control Synthesis for Robotics in the Presence of Temporal Logic Specifications
https://resolver.caltech.edu/CaltechTHESIS:12312015-131513787
Year: 2016
DOI: 10.7907/Z94Q7RW3
This thesis presents methods for incrementally constructing controllers in the presence of uncertainty and nonlinear dynamics. The basic setting is motion planning subject to temporal logic specifications. Broadly, two categories of problems are treated. The first is reactive formal synthesis when so-called discrete abstractions are available. The fragment of linear-time temporal logic (LTL) known as GR(1) is used to express assumptions about an adversarial environment and requirements of the controller. Two problems of changes to a specification are posed that concern the two major aspects of GR(1): safety and liveness. Algorithms providing incremental updates to strategies are presented as solutions. In support of these, an annotation of strategies is developed that facilitates repeated modifications. A variety of properties are proven about it, including necessity of existence and sufficiency for a strategy to be winning. The second category of problems considered is non-reactive (open-loop) synthesis in the absence of a discrete abstraction. Instead, the presented stochastic optimization methods directly construct a control input sequence that achieves low cost and satisfies a LTL formula. Several relaxations are considered as heuristics to address the rarity of sampling trajectories that satisfy an LTL formula and demonstrated to improve convergence rates for Dubins car and single-integrators subject to a recurrence task.https://resolver.caltech.edu/CaltechTHESIS:12312015-131513787Robustness, Adaptation, and Learning in Optimal Control
https://resolver.caltech.edu/CaltechTHESIS:06032016-102336160
Year: 2016
DOI: 10.7907/Z9F18WPB
<p>Recent technological advances have opened the door to a wide variety of dynamic control applications, which are enabled by increasing computational power in ever smaller devices. These advances are backed by reliable optimization algorithms that allow specification, synthesis, and embedded implementation of sophisticated learning-based controllers. However, as control systems become more pervasive, dynamic, and complex, the control algorithms governing them become more complex to design and analyze. In many cases, optimal control policies are practically impossible to determine unless the state dimension is small, or the dynamics are simple. Thus, in order to make implementation progress, the control designer must specialize to suboptimal architectures and approximate control. The major engineering challenge in the upcoming decades will be how to cope with the complexity of designing implementable control architectures for these smart systems while certifying their safety, robustness, and performance.</p>
<p>This thesis tackles the design and verification complexity by carefully employing tractable lower and upper bounds on the Lyapunov function, while making connections to robust control, formal synthesis, and machine learning. Specifically, optimization-based upper bounds are used to specify robust controllers, while lower bounds are used to obtain performance bounds and to synthesize approximately optimal policies. Implementation of these bounds depends critically on carrying out learning and optimization in the loop. Examples in aerospace, formal methods, hybrid systems, and networked adaptive systems are given, and novel sources of identifiability and persistence of excitation are discussed.</p>https://resolver.caltech.edu/CaltechTHESIS:06032016-102336160Two and Three Finger Caging of Polygons and Polyhedra
https://resolver.caltech.edu/CaltechTHESIS:12062015-164238181
Year: 2016
DOI: 10.7907/Z93X84KR
<p>Multi-finger caging offers a rigorous and robust approach to robot grasping. This thesis provides several novel algorithms for caging polygons and polyhedra in two and three dimensions. Caging refers to a robotic grasp that does not necessarily immobilize an object, but prevents it from escaping to infinity. The first algorithm considers caging a polygon in two dimensions using two point fingers. The second algorithm extends the first to three dimensions. The third algorithm considers caging a convex polygon in two dimensions using three point fingers, and considers robustness of this cage to variations in the relative positions of the fingers.</p>
<p>This thesis describes an algorithm for finding all two-finger cage formations of planar polygonal objects based on a contact-space formulation. It shows that two-finger cages have several useful properties in contact space. First, the critical points of the cage representation in the hand’s configuration space appear as critical points of the inter-finger distance function in contact space. Second, these critical points can be graphically characterized directly on the object’s boundary. Third, contact space admits a natural rectangular decomposition such that all critical points lie on the rectangle boundaries, and the sublevel sets of contact space and free space are topologically equivalent. These properties lead to a caging graph that can be readily constructed in contact space. Starting from a desired immobilizing grasp of a polygonal object, the caging graph is searched for the minimal, intermediate, and maximal caging regions surrounding the immobilizing grasp. An example constructed from real-world data illustrates and validates the method.</p>
<p>A second algorithm is developed for finding caging formations of a 3D polyhedron for two point fingers using a lower dimensional contact-space formulation. Results from the two-dimensional algorithm are extended to three dimension. Critical points of the inter-finger distance function are shown to be identical to the critical points of the cage. A decomposition of contact space into 4D regions having useful properties is demonstrated. A geometric analysis of the critical points of the inter-finger distance function results in a catalog of grasps in which the cages change topology, leading to a simple test to classify critical points. With these properties established, the search algorithm from the two-dimensional case may be applied to the three-dimensional problem. An implemented example demonstrates the method.</p>
<p>This thesis also presents a study of cages of convex polygonal objects using three point fingers. It considers a three-parameter model of the relative position of the fingers, which gives complete generality for three point fingers in the plane. It analyzes robustness of caging grasps to variations in the relative position of the fingers without breaking the cage. Using a simple decomposition of free space around the polygon, we present an algorithm which gives all caging placements of the fingers and a characterization of the robustness of these cages.</p>https://resolver.caltech.edu/CaltechTHESIS:12062015-164238181Electromyographic Signal Processing With Application To Spinal Cord Injury
https://resolver.caltech.edu/CaltechTHESIS:05312016-211459301
Year: 2016
DOI: 10.7907/Z9QJ7F99
<p>An Electromyogram or Electromyographic (EMG) signal is the recording of the electrical activity produced by muscles. It measures the electric currents generated in muscles during their contraction. The EMG signal provides insight into the neural activation and dynamics of the muscles, and is therefore important for many different applications, such as in clinical investigations that attempt to diagnose neuromuscular deficiencies. In particular, the work in this thesis is motivated by rehabilitation for patients with spinal cord injury. The EMG signal is very important for researchers and practitioners to monitor and evaluate the effect of the rehabilitation training and the condition of muscles, as the EMG signal provides information that helps infer the neural activity in the spinal cord. Before the work in this thesis, EMG analysis required significant amounts of manual labeling of interesting signal features. The motivation of this thesis is to fully automate the EMG analysis tasks and yield accurate, consistent results.</p>
<p>The EMG signal contains multiple muscle responses. The difficulty in processing the EMG signal arises from the fact that the transient muscle response is a transient signal with unknown arrival time, unknown duration, and unknown shape. In addition, the EMG signal recorded from patients with spinal cord injury during rehabilitation is very different from the EMG signal of normal healthy people undergoing the same motions. For example, some of the muscle responses are very weak and thus hard to detect. Because of this, general EMG processing tools and methods are either not applicable or insufficient.</p>
<p>The primary contribution of this thesis is the development of a wavelet-based, double-threshold algorithm for the detection of transient peaks in the EMG signal. The application of wavelet transform in the detection of transient signals has been studied extensively and employed successfully. However, most of the theories assume certain knowledge about the shapes of the transient signals, which makes it hard to be generalized to the transient signals with arbitrary shapes. The proposed detection scheme focuses on the more fundamental feature of most transient signals (in particular the EMG signal): peaks, instead of the shapes. The continuous wavelet transform with Mexican Hat wavelet is employed. This thesis theoretically derived a framework for selecting a set of scales based on the frequency domain information. Ridges are identified in the time-scale space to combine the wavelet coefficients from different scales. By imposing two thresholds, one on the wavelet coefficient and one on the ridge length, the proposed detection scheme can achieve both high recall and high precision. A systematic approach for selecting the optimal parameters via simulation is proposed and demonstrated. Comparing with other state-of-the-art detection methods, the proposed method in this thesis yields a better detection performance, especially in the low Signal-to-Noise-Ratio (SNR) environment.</p>
<p>Based on the transient peak detection result, the EMG signal is further segmented and classified into various groups of monosynaptic Motor Evoked Potentials (MEPs) and polysynaptic MEPs using techniques stemming from Principal Component Analysis (PCA), hierarchical clustering, and Gaussian mixture model (GMM). A theoretical framework is proposed to segment the EMG signal based on the detected peaks. The scale information of the detected peak is used to derive a measure for its effective support. Several different techniques have been adapted together to solve the clustering problem. An initial hierarchical clustering is first performed to obtain most of the monosynaptic MEPs. PCA is used to reduce the number of features and the effect of the noise. The reduced feature set is then fed to a GMM to further divide the MEPs into different groups of similar shapes. The method of breaking down a segment of multiple consecutive MEPs into individual MEPs is derived.</p>
<p>A software with graphic user interface has been implemented in Matlab. The software implements the proposed peak detection algorithm, and enables the physiologists to visualize the detection results and modify them if necessary. The solutions proposed in this thesis are not only helpful to the rehabilitation after spinal cord injury, but applicable to other general processing tasks on transient signals, especially on biological signals.</p>https://resolver.caltech.edu/CaltechTHESIS:05312016-211459301Wireless Nano and Molecular Scale Neural Interfacing
https://resolver.caltech.edu/CaltechTHESIS:12162016-094948729
Year: 2017
DOI: 10.7907/Z9RJ4GG4
Nanoscale circuits and sensors built from silicon nanowires, carbon nanotubes and other devices will require methods for unobtrusive interconnection with the macroscopic world to fully realise their potential; the size of conventional wires precludes their integration into dense, miniature systems. The same wiring problem presents an obstacle in our attempts to understand the brain by means of massively deployed nanodevices, for multiplexed recording and stimulation in vivo. We report on a nanoelectromechanical system that ameliorates wiring constraints, enabling highly integrated sensors to be read in parallel through a single output. Its basis is an effect in piezoelectric nanomechanical resonators that allows sensitive, linear and real-time transduction of electrical potentials. We interface multiple signals through a mechanical Fourier transform using tuneable resonators of different frequency and extract the signals from the system optically. With this method we demonstrate the direct transduction of neuronal action potentials from an extracellular microelectrode. We further extend this approach to incorporate nanophotonics for an all-optical system, coupled via a single optical fibre. Here, the mechanical resonators are both driven and probed optically, but modulated locally by the voltage sensors via the piezoelectric effect. Such piezophotonic nanoelectromechanical systems may be integrated with nanophotonic resonators, allowing concordant multiplexing in both the radiofrequency and optical bandwidths. In principle, this would allow billions of sensor channels to be multiplexed on an optical fibre. With view to eventually integrating such technology into a neural probe, we develop fabrication methods for crafting wired silicon neural probes via photolithography and electron beam lithography. Finally, to complement recording, we propose novel ideas for wireless, multiplexed neural stimulation through the use of radiofrequency-sensitive molecular scale resonators.https://resolver.caltech.edu/CaltechTHESIS:12162016-094948729Robotically Assembled Space Telescopes with Deployable Modules: Concepts and Design Methodologies
https://resolver.caltech.edu/CaltechTHESIS:07182016-100030713
Year: 2017
DOI: 10.7907/Z9T151NT
<p>This thesis first presents a novel architecture for robotically assembled optical telescopes with apertures between 20 m and 100 m, that utilizes only currently available technology. In this architecture, the primary mirror consists of two layers: a reflective layer and a truss backplane layer. The reflective layer is divided into mirror modules, or groups of mirror segments and actuators. The truss backplane layer is divided into truss modules that fold compactly for launch and are deployed in space by the robot. In this thesis, the design methodology of the mirror modules and truss modules is detailed. The ability of the designed truss layer to maintain precision requirements in the presence of typical space environment loads is demonstrated.</p>
<p>This architecture requires the deployment of many truss modules, and thus the deployment must be reliable despite errors introduced during manufacturing. In this thesis, a new simulation-based toolkit for estimating deployment reliability is described, including the experimental validation of the deployment simulation and the Monte Carlo-style method for repeating deployment simulations with different distributions of random fabrication errors to statistically estimate reliability. Using the toolkit, a set of reliability trade studies are then presented, revealing how different types of errors and design parameters affect reliability. Finally, the manufacturing tolerances and design modifications required to ensure high reliability are proposed.</p>
<p>Even if all modules deploy successfully, fabrication errors will still be present and may affect the assembly process. In this thesis, a new simulation method is presented that can model the step-by-step assembly of flexible modules with errors. The method is used to reveal that overall shape errors grow with the number of connections, resulting in significantly decreased surface precision and large-scale deformations from the nominal backplane shape as the size of the backplane increases. The misalignment at each individual connection does not increase as the backplane increases, but can still be much larger than the applied manufacturing tolerances simply due to random combinations. A simple design for the interconnects between modules is then tested, with simulation results demonstrating that it is unlikely to fully engage when the expected errors are present. With this information, a requirement on the complexity of the interconnect design is inferred, and potential modifications that may increase its efficacy are suggested.</p>https://resolver.caltech.edu/CaltechTHESIS:07182016-100030713Heading Estimation via Sun Sensing for Autonomous Navigation
https://resolver.caltech.edu/CaltechTHESIS:06142017-153929873
Year: 2017
DOI: 10.7907/Z9BG2M1S
<p>In preparation for the mission to Mars in 2020, NASA JPL and Caltech have been exploring the potential of sending a scout robot to accompany the new rover. One of the leading candidates for this scout robot is a lightweight helicopter that can fly every day for ~1 to 3 minutes. Its findings would be critical in the path planning for the rover because of its ability to see over and round local terrain elements. The inconsistent Mars magnetic field and GPS-denied environment would require the navigation system of such a vehicle to be completely overhauled. In this thesis, we present a novel technique for heading estimation for autonomous vehicles using sun sensing via fisheye camera. The approach results in accurate heading estimates within 2.4° when relying on the camera alone. If the information from the camera is fused with our sensors, the heading estimates are even more accurate. While this does not yet meet the desired error bound, it is a start with the critical flaws in the algorithm already identified in order to improve performance significantly. This lightweight solution however shows promise and does meet the weight constraints for the 1 kg Mars 2020 Helicopter Scout.</p>https://resolver.caltech.edu/CaltechTHESIS:06142017-153929873Online Learning for the Control of Human Standing via Spinal Cord Stimulation
https://resolver.caltech.edu/CaltechTHESIS:04172017-163725367
Year: 2017
DOI: 10.7907/Z9BK19DN
<p>Many applications in recommender systems or experimental design need to make decisions online. Each decision leads to a stochastic reward with initially unknown distribution, while new decisions are made based on the observations of previous rewards. To maximize the total reward, one needs to balance between exploring different strategies and exploiting currently optimal strategies within a given set of strategies. This is the underlying trade-off of a number of clinical neural engineering problems, including brain-computer interface, deep brain stimulation, and spinal cord injury therapy. In these systems, complex electronic and computational systems interact with the human central nervous system. A critical issue is how to control the agents to produce results which are optimal under some measure, for example, efficiently decoding the user's intention in a brain-computer interface or performs temporal and spatial specific stimulation in deep brain stimulation. This dissertation is motivated by electrical sipnal cord stimulation with high dimensional inputs(multi-electrode arrays). The stimulation is applied to promote the function and rehabilitation of the remaining neural circuitry below the spinal cord injury, and enable complex motor behaviors such as stepping and standing. To enable the careful tuning of these stimuli for each patient, the electrode arrays which deliver these stimuli have become increasingly more sophisticated, with a corresponding increase in the number of free parameters over which the stimuli need to be optimized. Since the number of stimuli is growing exponentially with the number of electrodes, algorithmic methods of selecting stimuli is necessary, particularly when the feedback is expensive to get.</p>
<p>In many online learning settings, particularly those that involve human feedback, reliable feedback is often limited to pairwise preferences instead of real valued feedback. Examples include implicit or subjective feedback for information retrieval and recommender systems, such as clicks on search results, and subjective feedback on the quality of recommended care. Sometimes with real valued feedback, we require that the sampled function values exceed some prespecified ``safety'' threshold, a requirement that existing algorithms fail to meet. Examples include medical applications where the patients' comfort must be guaranteed; recommender systems aiming to avoid user dissatisfaction; and robotic control, where one seeks to avoid controls that cause physical harm to the platform.</p>
<p>This dissertation provides online learning algorithms for several specific online decision-making problems. \selfsparring optimizes the cumulative reward with relative feedback. RankComparison deals with ranking feedback. \safeopt considers the optimization with real valued feedback and safety constraints. \cduel is designed for specific spinal cord injury therapy.</p>
<p>A variant of \cduel was implemented in closed-loop human experiments, controlling which epidural stimulating electrodes are used in the spinal cord of SCI patients. The results obtained are compared with concurrent stimulus tuning carried out by human experimenter. These experiments show that this algorithm is at least as effective as the human experimenter, suggesting that this algorithm can be applied to the more challenging problems of enabling and optimizing complex, sensory-dependent behaviors, such as stepping and standing in SCI patients.</p>
<p>In order to get reliable quantitative measurements besides comparisons, the standing behaviors of paralyzed patients under spinal cord stimulation are evaluated. The potential of quantifying the quality of bipedal standing in an automatic approach is also shown in this work.</p>https://resolver.caltech.edu/CaltechTHESIS:04172017-163725367Optimal Sensor Placement for Bayesian Parametric Identification of Structures
https://resolver.caltech.edu/CaltechTHESIS:05182017-090742614
Year: 2017
DOI: 10.7907/Z9H41PG5
<p>There exists a choice in where to place sensors to collect data for Bayesian model updating and system identification of structures. It is desirable to use an available deterministic predictive model, such as a finite-element model, along with prior information on the uncertain model parameters and the uncertain accuracy of the predictive model, to determine which optimal sensor locations should be instrumented in the structure. In this thesis, an information-theoretic framework for optimality is considered.</p>
<p>The mutual information between the uncertain model predictions for the data and the uncertain model parameters is presented as a natural measure of reduction in uncertainty to maximize over sensor configurations. A combinatorial search over all sensor configurations is usually prohibitively expensive. A convex optimization method is developed to provide a fast sub-optimal, but possibly optimal, sensor configuration when certain simplifying assumptions can be made about the chosen stochastic model class for the structure. The optimization method is demonstrated to work for a 50-story uniform shear building, with 20 sensors to be installed.</p>
<p>The stability of optimal sensor configurations under refinement of the mesh of the underlying finite-element model is investigated and related to the choice of prediction-error correlations in the model. An example problem of placement of a single sensor on the continuum of an elastic axial bar is solved analytically.</p>
<p>In order to solve the optimal sensor placement problem in the more general case, numerical estimation of mutual information between the model predictions for the data and the model parameters becomes necessary. To this end, a thermodynamic integration scheme based on path sampling is developed with the aim of estimating the entropy of the data prediction distribution. The scheme is demonstrated to work for an example that uses synthetic data for model class comparison between linear and Duffing oscillator model classes. The thermodynamic integration method is then used to determine the optimal location of a single sensor for a two degree-of-freedom oscillator model.</p>
https://resolver.caltech.edu/CaltechTHESIS:05182017-090742614Three-Dimensional Nano-Architected Materials as Platforms for Designing Effective Bone Implants
https://resolver.caltech.edu/CaltechTHESIS:12242017-060345135
Year: 2018
DOI: 10.7907/Z947482K
<p>The growing world population coupled with longer human life expectancy warrants the need for better medical implant development. Recent advances in lithographic techniques have opened the door to a variety of approaches to tackle the aforementioned issue. However, several scientific hurdles must be overcome before patients can use fully synthetic and effective implants.</p>
<p>Identifying the optimal material, porosity, and mechanical properties of the scaffold to induce cell functionality are key obstacles. Limitations in established fabrication techniques have hindered the ability to fully understand cell behavior on 3D substrates. 3D printing is limited to feature sizes that are at least one order of magnitude larger than a single cell (~10μm); electrospinning is able to yield features that are on the same scale as cells, but its stochastic nature leads to scaffolds with poor mechanical properties; salt leeching doesn’t allow for control of pore size and distribution which have detrimental effects on nutrient diffusion and cell ingrowth, thereby thwarting the formation of functional tissue.</p>
<p>Much effort has been made to create a suitable platform for regenerating a relatively less complex organ, such as bone, yet the inability to fully understand cell mechanics on 3D scaffolds has curbed the fabrication of effective bone implants.</p>
<p>The first part of this thesis focuses on the suitability of nanoarchitected materials as 3D platforms for bone-tissue growth. We employed two-photon lithography to create polymeric and hydroxyapatite-coated 3D nanolattices to explore scaffold biocompatibility and material effects on osteoblast attachment and growth. Our experiments showed that the unit cell geometry, tetrakaidekahedron, and size, 25μm, were adequate for cell attachment and infiltration, which are hallmark signs of biocompatibility. Our study also corroborated previous findings that mammalian cells respond differently to different materials that they come in contact with. To isolate structural effects, we fabricated nanolattices coated with a uniform 20nm-thick outermost layer of TiO<sub>2</sub>. These nanolattices, which had fixed porosity and unit cell size (25μm) while they varied in structural stiffness (~2-9MPa) were used to explore the influence of scaffold properties on the viability of osteoblasts in a microenvironment similar to that of natural bone. Upon growing osteogenic cells on the nanolattices, significant cell attachment and presence of various calcium phosphate species, which are commonly found in natural bone, were observed. These findings suggest that 3-dimensional nano-architected materials can be used as effective scaffolds for bone cell growth and proliferation.</p>
<p>The second part of the thesis investigates the effects of nanolattice structural stiffness and loading conditions on osteoblast behavior. We fabricated nanolattices with stiffness ranging from ~0.7MPa to 100MPa. Experiments done by seeding osteoblast-like cells on these nanolattices revealed that both stress fiber concentration and bioapatite deposition were higher on the most compliant nanolattice, (0.7 MPa) by ~20% and ~40% respectively. These results provide insights into cell behavior in 3D microenvironments which can lead to a better understanding of stress shielding at the cellular level. Preventing stress shielding by creating scaffolds with structural stiffness and porosity that enhances osteoblasts activity could lead to the creation of effective implants with improved mechanical stability which ultimately improves osteointegration.</p>
<p>In addition to investigating static cell-scaffold interactions we took advantage of the nanolattices tunability to study the effects of dynamic loading on cell behavior. Bone adaptation is driven by dynamic, rather than static loading, however there is still wide controversy on whether stress, strain or loading frequency plays the most significant role in bone remodeling, which drives bone healing.</p>
<p>In order to understand cell sensitivity to varying loads, displacements and frequencies, we fabricated hollow TiO<sub>2</sub> nanolattices with stiffness ranging from ~0.7-35MPa which were populated with osteoblast-like cells and subjected to cyclic compression to either a constant stress or strain. After seeding SAOS-2 cells on the nanolattices for 12 days different dynamic loading conditions were tested: (1) cyclic uniaxial compressions to strains ranging from ~0.3-2% strain were carried out to investigate the effects of strain magnitude on cell behavior. (2) Cyclic uniaxial compressions to stresses spanning from ~0.02-1MPa were performed to explore the role of stress magnitude on the cells’ stress fibers formation. (3) The nanolattices were cyclically loaded at different frequencies, ~0.1-3Hz, while maintaining stress and strain constant, which provided insights into how loading frequency affects osteoblasts behavior.</p>
<p>Cell activity, which was measured by monitoring f-actin and vinculin fluorescence intensity, revealed increased fluorescence in those cells that were mechanically stimulated as opposed to those that were statically grown on the nanolattices regardless of loading condition. Cell response was most drastically affected by varying the loading frequency. A ~30% increase in f-actin fluorescence was observed in the cells grown on the nanolattices that were loaded at ~3Hz compared to those that were grown on the nanolattices that were cyclically compressed at ~0.1Hz.</p>
<p>The last part of this thesis is focused on developing a three-dimensional architected capacitor that could be used as a strain gauge to further our understanding of cell mechanics in 3D. We took advantage of the mechanical tunability of the nanolattices to fabricate a 3D parallel-plate capacitor with a basal capacitance of ~280fF and able to sense forces as low as ~30μN. This work points to nano-architected materials as promising candidates for ideal platforms to investigate more realistic cellular conditions which can immensely benefit the field of tissue engineering.</p>https://resolver.caltech.edu/CaltechTHESIS:12242017-060345135Dynamic Modeling and Control of Spherical Robots
https://resolver.caltech.edu/CaltechTHESIS:05302018-110559204
Year: 2018
DOI: 10.7907/E5CW-8H41
<p>In this work, a rigorous framework is developed for the modeling and control of spherical robotic vehicles. Motivation for this work stems from the development of Moball, which is a self-propelled sensor platform that harvests kinetic energy from local wind fields. To study Moball's dynamics, the processes of Lagrangian reduction and reconstruction are extended to robotic systems with symmetry-breaking potential energies, in order to simplify the resulting dynamic equations and expose mathematical structures that play an important role in subsequent control-theoretic tasks. These results apply to robotic systems beyond spherical robots. A formulaic procedure is introduced to derive the reduced equations of motion of most spherical robots from inspection of the Lagrangian. This adaptable procedure is applied to a diverse set of robotic systems, including multirotor aerial vehicles.</p>
<p>Small time local controllability (STLC) results are derived for barycentric spherical robots (BSR), which are spherical vehicles whose locomotion depends on actuating the vehicle's center of mass (COM) location. STLC theorems are introduced for an arbitrary BSR on flat, sloped, or smooth terrain. I show that STLC depends on the surjectivity of a simple <i>steering matrix</i>. An STLC theorem is also derived for a class of commonly encountered multirotor vehicles.</p>
<p>Feedback linearizing and PID controllers are proposed to stabilize an arbitrary spherical robot to a desired trajectory over smooth terrain, and direct collocation is used to develop a feedforward controller for Moball specifically. Moball's COM is manipulated by a novel system of magnets and solenoids, which are actuated by a "ballistic-impulse" controller that is also presented. Lastly, a motion planner is developed for energy-harvesting vehicles. This planner charts a path over smooth terrain while balancing the desire to achieve scientific objectives, avoid hazards, and the imperative of exposing the vehicle to environmental sources of energy such as local wind fields and topology. Moball's design details and experimental results establishing Moball's energy-harvesting performance (7<i>W</i> while rolling at a speed of 2 <i>m/s</i>), are contained in an Appendix.</p>https://resolver.caltech.edu/CaltechTHESIS:05302018-110559204An Electrophysiological Study Of Voluntary Movement and Spinal Cord Injury
https://resolver.caltech.edu/CaltechTHESIS:06012018-140912331
Year: 2018
DOI: 10.7907/K6P2-ZH75
<p>Voluntary movement is generated from the interaction between neurons in our brain and the neurons in our spinal cord that engage our muscles. A spinal cord injury destroys the connection between these two regions, but parts of their underlying neural circuits survive. A new class of treatment (the brain-machine interface) takes advantage of this fact by either a) recording neural activity from the brain and predicting the intended movement (neural prosthetics) or b) stimulating neural activity in the spinal cord to facilitate muscle activity (spinal stimulation). This thesis covers new research studying the brain-machine interface and its application for spinal injury.</p>
<p>First, the electrical properties of the microelectrode (the main tool of the brain-machine interface) are studied during deep brain recording and stimulation. This work shows that the insulation coating the electrode forms a capacitor with the surrounding neural tissue. This capacitance causes large spikes of voltage in the surrounding tissue during deep brain stimulation, which will cause electrical artifacts in neural recordings and may damage the surrounding neurons. This work also shows that a coaxially shielded electrode will block this effect.</p>
<p>Second, the activity of neurons in the parietal cortex is studied during hand movements, which has applications for neural prosthetics. Prior work suggests that the parietal cortex encodes a state-estimator [1], which combines sensory feedback with the internal efference copy to predict the state of the hand. To test this idea, we used a visual lag to misalign sensory feedback from the efference copy. The expectation was that a state-estimator would unknowingly combine the delayed visual feedback with the current efference information, resulting in incorrect predictions of the hand. Our results show a drop in correlation between neural activity in the parietal cortex and hand movement during a visual lag, supporting the idea that the parietal cortex encodes a state-estimator. This correlation gradually recovers over time, showing that parietal cortex is adaptive to sensory delays.</p>
<p>Third, while the intention of spinal stimulation was to interact locally with neural circuits in the spinal cord, results from the clinic show that electrical stimulation of the lumbosacral enlargement enables paraplegic patients to regain voluntary movement of their legs [2]. This means that spinal stimulation facilitates communication across an injury site. To further study this effect, we developed a new behavioral task in the rodent. Rats were trained to kick their right hindlimb in response to an auditory cue. The animals then received a spinal injury that caused paraplegia. After injury, the animals recovered the behavior (they could kick in response to the cue), but only during spinal stimulation. Their recovered behavior was slower and more stereotyped than their pre-injury response. Administering quipazine to these rodents disrupted their ability to respond to the cue, suggesting that serotonin plays an important role in the recovered pathway. This work proves that the new behavioral task is a successful tool for studying the recovery of voluntary movement.</p>
<p>Future work will combine cortical recordings with this behavioral task in the rodent to study plasticity in the nervous system and improve treatment of spinal cord injuries.</p>
<p>[1] Mulliken, Grant H., Sam Musallam, and Richard A. Andersen. "Forward estimation of movement state in posterior parietal cortex." Proceedings of the National Academy of Sciences105.24 (2008): 8170-8177.</p>
<p>[2] Harkema, Susan, et al. "Effect of epidural stimulation of the lumbosacral spinal cord on voluntary movement, standing, and assisted stepping after motor complete paraplegia: a case study." The Lancet 377.9781 (2011): 1938-1947.</p>https://resolver.caltech.edu/CaltechTHESIS:06012018-140912331Optimal Controller Synthesis for Nonlinear Systems
https://resolver.caltech.edu/CaltechTHESIS:12162017-121220572
Year: 2018
DOI: 10.7907/Z9TX3CK8
<p>Optimal controller synthesis is a challenging problem to solve. However, in many applications such as robotics, nonlinearity is unavoidable. Apart from optimality, correctness of the system behaviors with respect to system specifications such as stability and obstacle avoidance is vital for engineering applications. Many existing techniques consider either the optimality or the correctness of system behavior. Rarely, a tool exists that considers both. Furthermore, most existing optimal controller synthesis techniques are not scalable because they either require ad-hoc design or they suffer from the curse of dimensionality.</p>
<p>This thesis aims to close these gaps by proposing optimal controller synthesis techniques for two classes of nonlinear systems: linearly solvable nonlinear systems and hybrid nonlinear systems. Linearly solvable systems have associated Hamilton- Jacobi-Bellman (HJB) equations that can be transformed from the original nonlinear partial differential equation (PDE) into a linear PDE through a logarithmic transformation. The first part of this thesis presets two methods to synthesize optimal controller for linearly solvable nonlinear systems. The first technique uses a hierarchy of sums-of-square programs to compute a sequence of suboptimal controllers that have non-increasing suboptimality for first exit and finite horizon problems. This technique is the first systematic approach to provide stability and suboptimal performance guarantees for stochastic nonlinear systems in one framework. The second technique uses the low rank tensor decomposition framework to solve the linear HJB equation for first exit, finite horizon, and infinite horizon problems. This technique scale linearly with dimensions, alleviating the curse of dimensionality and enabling us to solve the linear HJB equation for a quadcopter model that is a twelve-dimensional system on a personal laptop. A new algorithm is proposed for a key step in the controller synthesis algorithm to solve the ill-conditioning issue that arises in the original algorithm. A MATLAB toolbox that implements the algorithms is developed, and the performance of these algorithms is illustrated by a few engineering examples.</p>
<p>Apart from stability, in many applications, more complex specifications such as obstacle avoidance, reachability, and surveillance are required. The second part of the thesis describes methods to synthesize optimal controllers for hybrid nonlinear systems with quantitative objectives (i.e., minimizing cost) and qualitative objectives (i.e., satisfying specifications). This thesis focuses on two types of qualitative objectives, regular objectives, and ω-regular objectives. Regular objectives capture bounded time behavior such as reachability, and ω-regular objectives capture long term behavior such as surveillance. For both types of objectives, an abstraction-refinement procedure that preserves the cost is developed. A two-player game is solved on the product of the abstract system and the given objectives to synthesize the suboptimal controller for the hybrid nonlinear system. By refining the abstract system, the algorithms are guaranteed to converge to the optimal cost and return the optimal controller if the original systems are robust with respect to the initial states and the optimal controller inputs. The proposed technique is the first abstraction-refinement based technique to combine both quantitative and qualitative objectives into one framework. A Python implementation of the algorithms are developed, and a few engineering examples are presented to illustrate the performance of these algorithms.</p>https://resolver.caltech.edu/CaltechTHESIS:12162017-121220572Partially Mixed Selectivity and Parietal Cortex
https://resolver.caltech.edu/CaltechTHESIS:05212018-145923990
Year: 2018
DOI: 10.7907/R1RS-RJ59
<p>Brain-machine interfaces (BMIs) decode intention signals and other variables from the brain in order to control a computer, tablet, or prosthetic limb. In order to improve the technology, a better understanding of the representational mechanisms within the brain is necessary. Here we study how the anterior intraparietal area (AIP) of human posterior parietal cortex is able to represent many variables within a small patch of cortex. We record single unit activity using a 4 x 4 mm microelectrode array implanted in AIP of a human tetraplegic volunteer. Testing movements of different cognitive strategies, body parts, and body sides, we find that the neural population represents information in a high-dimensional way, termed "mixed selectivity", with individual units coding for idiosyncratic combinations of variables. Furthermore, we find that the variables are not randomly mixed but exhibited "partially mixed selectivity" with certain variables more randomly mixed than others. Representations were "functionally segregated", with representations of the hand and shoulder largely orthogonal despite the high degree of anatomical overlap; representations of body side and strategy were organized by body part. We also examine how the representations changed between BMI training and online BMI control. We find that the structure of the movement representations was preserved, with the different representations found during calibration maintained during online control. Finally, we study the sensory mirror system, a system that processes observed sensations similarly to experienced sensations. We once again find partially mixed selectivity and functional segregation by body parts, showing that this method of encoding information exists not just in the action intention domain but also in the sensory domain. Our results propose partially mixed selectivity as a general mechanism for encoding high dimensional in formation in a small neural population, while also advancing the possibility of limited electrode-array BMIs decoding movements of a large extent of the body.</p> https://resolver.caltech.edu/CaltechTHESIS:05212018-145923990Electromechanical Properties of 3D Multifunctional Nano-Architected Materials
https://resolver.caltech.edu/CaltechTHESIS:01182019-105653047
Year: 2019
DOI: 10.7907/D0AD-4T88
<p>In this thesis, we explore the fabrication and characterization of 3D architected multifunctional materials in three different categories: varied density for tailored mechanical response, stiff ultra low-<i>k</i> dielectric materials, and direct laser writing of piezoelectric structures at the micron scale. The density of an architected material plays a large role in determining its effective Young’s modulus, strength, and deformation behavior. The first section of this work explores the effect of incorporating two density regions into hollow nanolattices, which results in two distinct mechanical response regions for horizontal interfaces and a combined varying response for a diagonal interface. The second section of this work describes low dielectric constant (low-<i>k</i>) materials, which have gained increasing popularity because of their critical role in developing faster, smaller, and higher performance devices. We report the fabrication of 3D nanoarchitected hollow-beam alumina dielectrics with a <i>k</i> value of 1.06 - 1.10 at 1 MHz that is stable over the voltage range of -20 to 20 V and a frequency range of 100 kHz to 10 MHz, with an effective Young’s modulus of 30 MPa, a strength of 1.07 MPa, a nearly full shape recoverability to its original size after >50% compressions, and outstanding thermal stability with a thermal coefficient of dielectric constant (TCK) of 2.43 x 10<sup>-5</sup>K<sup>-1</sup> up to 800° C. Finally, we report the fabrication of monolithic piezoelectric ZnO structures of arbitrary shape via a polymer complex route. We have confirmed the microstructure using XRD, TEM, and SAED, and have observed its electromechanical response using a novel in-situ experiment.</p>https://resolver.caltech.edu/CaltechTHESIS:01182019-105653047Sparse Deconvolution with Applications to Spike Sorting
https://resolver.caltech.edu/CaltechTHESIS:04292019-161916232
Year: 2019
DOI: 10.7907/3JQS-BT21
<p>Chronic extracellular recording is the use of implanted electrodes to measure the electrical activity of nearby neurons over a long period of time. It presents an unparalleled view of neural activity over a broad range of time scales, offering sub-millisecond resolution of single action potentials while also allowing for continuous recording over the course of many months. These recordings pick up a rich collection of neural phenomena -- spikes, ripples, and theta oscillations, to name a few -- that can elucidate the activity of individual neurons and local circuits.</p>
<p>However, this also presents an interesting challenge for data analysis. Chronic extracellular recordings contain overlapping signals from multiple sources, requiring these signals to be detected and classified before they can be properly analyzed. The combination of fine temporal resolution with long recording durations produces large datasets, requiring efficient algorithms that can operate at scale.</p>
<p>In this thesis, I consider the problem of spike sorting: detecting spikes (the extracellular signatures of individual neurons' action potentials) and clustering them according to their putative source. First, I introduce a sparse deconvolution approach to spike detection, which seeks to detect spikes and represent them as the linear combination of basis waveforms. This approach is able to separate overlapping spikes without the need for source templates, and produces an output that can be used with a variety of clustering algorithms.</p>
<p>Second, I introduce a clustering algorithm based around a mixture of drifting <i>t</i>-distributions. This model captures two features of chronic extracellular recordings -- cluster drift over time and heavy-tailed residuals in the distribution of spikes -- that are missing from previous models. This enables us to reliably track individual neurons over longer periods of time. I will also show that this model produces more accurate estimates of classification error, which is an important component to proper interpretation of the spike sorting output.</p>
<p>Finally, I present a few theoretical results that may assist in the efficient implementation of sparse deconvolution.</p>https://resolver.caltech.edu/CaltechTHESIS:04292019-161916232Representations of Action Monitoring and Cognitive Control by Single Neurons in the Human Brain
https://resolver.caltech.edu/CaltechTHESIS:06012019-234115726
Year: 2019
DOI: 10.7907/RG37-G744
Cognitive control arises whenever a prepotent and often automatic response needs to be overcome by another response. Control is usually effortful and relies on monitoring processes that detect when control is needed and/or when it failed. Control is one of the most important aspects of human behavior in everyday life and is a critical component of executive function. In a series of three empirical chapters, I present results from invasive single-neuron recordings from the frontal cortex of neurosurgical human patients while they perform tasks requiring cognitive control. I show that a substantial proportion of neurons in the pre-supplementary motor area (pre-SMA), and in the dorsal anterior cingulate cortex (dACC), signal response errors shortly after they occurred, but well before onset of feedback. Here I demonstrate that these error neurons signal self-detected errors and that they were separate from neurons signaling conflict. The response of error neurons correlated trial-by-trial with the simultaneously recorded intracranial error-related negativity (iERN), thereby establishing a single-neuron correlate of this important scalp potential. iERN-error neuron synchrony in dACC, but not pre-SMA, predicted whether post-error slowing, which is a measure of control, occurred or not. Spike-field coherence between action potentials and local field potentials in specific frequency bands, and latency differences between the different brain regions, suggest a mechanistic model whereby information relevant to control is passed between sectors of the medial frontal cortex. Multiplexing of different ex-post monitoring signals by individual neurons further documents that control relies on multiple sources of information, which can be dynamically routed in the brain depending on task demands. These findings provide the most complete set of single-neuron data on how errors and conflict signals at the single neuron level contribute to cognitive controls in humans. They provide a first-single neuron correlate of an extensively utilized scalp EEG potential. Together, this work provides a strong complement to investigations of this topic using fMRI in humans, and using electrophysiology in monkeys, and suggests specific future directions.https://resolver.caltech.edu/CaltechTHESIS:06012019-234115726Decomposing Formal Specifications Into Assume-Guarantee Contracts for Hierarchical System Design
https://resolver.caltech.edu/CaltechTHESIS:07202018-115217471
Year: 2019
DOI: 10.7907/Z9Q52MTD
<p>Specifications for complex engineering systems are typically decomposed into specifications for individual subsystems in a way that ensures they are implementable and simpler to develop further. We describe a method to algorithmically construct specifications for components that should implement a given specification when assembled. By eliminating variables that are irrelevant to realizability of each component, we simplify the specifications and reduce the amount of information necessary for operation.
To identify these variables, we parametrize the information flow between components.</p>
<p>The specifications are written in the Temporal Logic of Actions, TLA+, with liveness properties restricted to an implication of conjoined recurrence properties, known as GR(1). We study whether GR(1) contracts exist in the presence of full information, and prove that memoryless GR(1) contracts that preserve safety do not always exist, whereas contracts in GR(1) with history-determined variables added do exist. We observe that timed stutter-invariant specifications of open-systems in general require GR(2) liveness properties for expressing them.</p>
<p>We formalize a definition of realizability in TLA+, and define an operator for forming open-systems from closed-systems, based on a variant of the while-plus operator. The resulting open-system properties are realizable when expected to be. We compare stepwise implication operators from the literature, and establish relations between them, and examine the arity required for expressing these operators. We examine which symmetric combinations of stepwise implication and implementation kind avoid circular dependence, and show that only Moore components specified by strictly causal stepwise implication avoid circular dependence.</p>
<p>The proposed approach relies on symbolic algorithms for computing specifications. To convert the generated specifications from binary decision diagrams to readable formulas over integer variables, we symbolically solve a minimal covering problem. We implemented an algorithm for minimal covering over lattices originally proposed for two-level logic minimization. We formalized the computation of essential elements and cyclic core that is part of this algorithm, and machine-checked the proofs of safety properties using a proof assistant. Proofs supporting the thesis are organized as TLA+ modules in appendices.</p>https://resolver.caltech.edu/CaltechTHESIS:07202018-115217471Numerical Investigation of Spinal Neuron Facilitation with Multi-electrode Epidural Stimulation
https://resolver.caltech.edu/CaltechTHESIS:11302018-185025297
Year: 2019
DOI: 10.7907/2DVK-G212
<p>Approximately 1,275,000 people in the US have a spinal cord injury severe enough to cause some paralysis of the arms and/or legs. Epidural stimulation using implanted multi-electrode stimulating arrays over the lumbosacral spinal cord has recently shown promise in assisting individuals with severe spinal cord injuries to stand, walk, and even facilitate voluntary movement. Both animal model and human studies have shown that sub-threshold facilitation of motor recovery gives the best results. The underlying neural mechanisms by which sub-threshold epidural stimulation leads to motor recovery are incompletely known.</p>
<p>This thesis uses computational methods to study the <i>facilitation effect</i>. A neuron is facilitated if a sub-threshold synaptic input can cause a neuronal output under the influence of a stimulating electric field. The analysis in this thesis is based on a computational model of the epidural spinal stimulation process in the rat spinal cord. This model includes a time-domain finite element simulation (using COMSOL®) of the various tissues in the spinal cord with the appropriate anisotropic and frequency-dependent complex relative permittivities. The voltages obtained from the finite element simulations were used as the extracellular voltage in NEURON simulations.</p>
<p>A population of neurons were simulated under a wide variety of conditions. These simulations highlight the effect of neuron orientation, location, and synaptic timing as key parameters which influence facilitation.</p>
<p>This study indicates that regions of the spinal cord that have previously been ignored may be actively involved in motor recovery. These results may also enable the design of specialized epidural electrode arrays and the design of new stimulation protocols.</p>https://resolver.caltech.edu/CaltechTHESIS:11302018-185025297Towards High Performance Robotic Actuation
https://resolver.caltech.edu/CaltechTHESIS:05222019-132217207
Year: 2019
DOI: 10.7907/W64Q-1R69
<p>The main objective of this thesis is to enable development of high performance actuation for legged, limbed and mobile robots. Due to the fact that such robots need to support their own weight, their actuators need to be light weight, compact and efficient. Furthermore, a dynamics analysis, shows that the actuators' design may have significant impact on a robot's dynamics sensitivity. These consideration motivate improvements in all actuator design aspects.</p>
<p>First, the application-specific design of outer rotor motors with concentrated windings is considered. It is shown that an intrinsic design trade-off exists between a motor's copper loss, core loss and mass, which allows development of motors with superior performance for a particular application. The three main application categories of interest are: electric vehicles, drones and robotic joints. Due to their outstanding torque density, high pole count outer rotor motors are analysed in terms of their design and optimization for robotic applications. Motor design scaling modes are also described in order to outline the main challenges in the implementation of high torque motors.</p>
<p>Next, the design of gearboxes for robotic actuation is discussed. A novel type of high reduction Bearingless Planetary Gearbox is introduced which allows large range of reduction ratios to be achieved in a compound planetary stage. In this concept, all gear components float in an unconstrained manner as the planet carrier is substituted with a secondary sun gear. The advantages of the Bearingless Planetary Gearbox over current approaches in terms of improved robustness, load distribution, manufacturability, and assembly are outlined.</p>
<p>Finally, analysis, design, and prototyping of rotary planar springs for rotary series elastic actuators is described. A mathematical model, based on curved beam theory, that allows rapid design, analysis, and comparison of rotary springs is developed. Mass reduction techniques based on composite arm structures are introduced and internal arm contact modeling is presented. Motivated by strain energy density analysis, an optimization based spring design approach is developed that allows significant increase in the torque and torque density.</p>
https://resolver.caltech.edu/CaltechTHESIS:05222019-132217207Connecting the Speed-Accuracy Trade-Offs in Sensorimotor Control and Neurophysiology Reveals Diversity Sweet Spots in Layered Control Architectures
https://resolver.caltech.edu/CaltechTHESIS:06072019-083145024
Year: 2019
DOI: 10.7907/VYQY-DF47
Nervous systems sense, communicate, compute, and actuate movement using distributed components with trade-offs in speed, accuracy, sparsity, noise, and saturation. Nevertheless, the resulting control can achieve remarkably fast, accurate, and robust performance due to a highly effective layered control architecture. However, this architecture has received little attention from the existing research. This is in part because of the lack of theory that connects speed-accuracy trade-offs (SATs) in the components neurophysiology with system-level sensorimotor control and characterizes the overall system performance when different layers (planning vs. reflex layer) act work jointly. In thesis, we present a theoretical framework that provides a synthetic perspective of both levels and layers. We then use this framework to clarify the properties of effective layered architectures and explain why there exists extreme diversity across layers (planning vs. reflex layers) and within levels (sensorimotor versus neural/muscle hardware levels). The framework characterizes how the sensorimotor SATs are constrained by the component SATs of neurons communicating with spikes and their sensory and muscle endpoints, in both stochastic and deterministic models. The theoretical predictions are also verified using driving experiments. Our results lead to a novel concept, termed ``diversity sweet spots (DSSs)'': the appropriate diversity in the properties of neurons and muscles across layers and within levels help create systems that are both fast and accurate despite being built from components that are individually slow or inaccurate. At the component level, this concept explains why there are extreme heterogeneities in the neural or muscle composition. At the system level, DSSs explain the benefits of layering to allow extreme heterogeneities in speed and accuracy in different sensorimotor loops. Similar issues and properties also extend down to the cellular level in biology and outward to our most advanced network technologies from smart grid to the Internet of Things. We present our initial step in expanding our framework to that area and widely-open area of research for future direction. https://resolver.caltech.edu/CaltechTHESIS:06072019-083145024Functional Autonomy Techniques for Manipulation in Uncertain Environments
https://resolver.caltech.edu/CaltechTHESIS:06082020-104419929
Year: 2020
DOI: 10.7907/0kgt-yg76
<p>As robotic platforms are put to work in an ever more diverse array of environments, their ability to deploy visuomotor capabilities without supervision is complicated by the potential for unforeseen operating conditions. This is a particular challenge within the domain of manipulation, where significant geometric, semantic, and kinetic understanding across the space of possible manipulands is necessary to allow effective interaction. To facilitate adoption of robotic platforms in such environments, this work investigates the application of functional, or behavior level, autonomy to the task of manipulation in uncertain environments. Three functional autonomy techniques are presented to address subproblems within the domain.</p>
<p>The task of reactive selection between a set of actions that incur a probabilistic cost to advance the same goal metric in the presence of an operator action preference is formulated as the Obedient Multi-Armed Bandit (OMAB) problem, under the purview of Reinforcement Learning. A policy for the problem is presented and evaluated against a novel performance metric, disappointment (analogous to prototypical MAB's regret), in comparison to adaptations of existing MAB policies. This is posed for both stationary and non-stationary cost distributions, within the context of two example planetary exploration applications of multi-modal mobility, and surface excavation.</p>
<p>Second, a computational model that derives semantic meaning from the outcome of manipulation tasks is developed, which leverages physics simulation and clustering to learn symbolic failure modes. A deep network extracts visual signatures for each mode that may then guide failure recovery. The model is demonstrated through application to the archetypal manipulation task of placing objects into a container, as well as stacking of cuboids, and evaluated against both synthetic verification sets and real depth images.</p>
<p>Third, an approach is presented for visual estimation of the minimum magnitude grasping wrench necessary to extract massive objects from an unstructured pile, subject to a given end effector's grasping limits, that is formulated for each object as a "wrench space stiction manifold". Properties are estimated from segmented RGBD point clouds, and a geometric adjacency graph used to infer incident wrenches upon each object, allowing candidate extraction object/force-vector pairs to be selected from the pile that are likely to be within the system's capability.</p>https://resolver.caltech.edu/CaltechTHESIS:06082020-104419929Applied Safety Critical Control
https://resolver.caltech.edu/CaltechTHESIS:06022020-154234707
Year: 2020
DOI: 10.7907/y97v-b205
<p>There is currently a clear gap between control-theoretical results and the reality of robotic implementation, in the sense that it is very difficult to transfer analytical guarantees to practical ones. This is especially problematic when trying to design safety-critical systems where failure is not an option. While there is a vast body of work on safety and reliability in control theory, very little of it is actually used in practice where safety margins are typically empiric and/or heuristic. Nevertheless, it is still widely accepted that a solution to these problems can only emerge from rigorous analysis, mathematics, and methods. In this work, we therefore seek to help bridge this gap by revisiting and expanding existing theoretical results in light of the complexity of hardware implementation.</p>
<p>To that end, we begin by making a clear theoretical distinction between systems and models, and outline how the two need to be related for guarantees to transfer from the latter to the former. We then formalize various imperfections of reality that need to be accounted for at a model level to provide theoretical results with better applicability. We then discuss the reality of digital controller implementation and present the mathematical constraints that theoretical control laws must satisfy for them to be implementable on real hardware. In light of these discussions, we derive new realizable set-invariance conditions that, if properly enforced, can guarantee safety with an arbitrary high levels of confidence. We then discuss how these conditions can be rigorously enforced in a systematic and minimally invasive way through convex optimization-based Safety Filters. Multiple safety filter formulations are proposed with varying levels of complexity and applicability. To enable the use of these safety filters, a new algorithm is presented to compute appropriate control invariant sets and guarantee feasibility of the optimization problem defining these filters. The effectiveness of this approach is demonstrated in simulation on a nonlinear inverted pendulum and experimentally on a simple vehicle. The aptitude of the framework to handle a system's dynamics uncertainty is illustrated by varying the mass of the vehicle and showcasing when safety is conserved. Then, the aptitude of this approach to provide guarantees that account for controller implementation's constraints is illustrated by varying the frequency of the control loop and again showcasing when safety is conserved.</p>
<p>In the second part of this work, we revisit the safety filtering approach in a way that addresses the scalability issues of the first part of this work. There are two main approaches to safety-critical control. The first one relies on computation of control invariant sets and was presented in the first part of this work. The second approach draws from the topic of optimal control and relies on the ability to realize Model-Predictive-Controllers online to guarantee the safety of a system. In that online approach, safety is ensured at a planning stage by solving the control problem subject for some explicitly defined constraints on the state and control input. Both approaches have distinct advantages but also major drawbacks that hinder their practical effectiveness, namely scalability for the first one and computational complexity for the second one. We therefore present an approach that draws from the advantages of both approaches to deliver efficient and scalable methods of ensuring safety for nonlinear dynamical systems. In particular, we show that identifying a backup control law that stabilizes the system is in fact sufficient to exploit some of the set-invariance conditions presented in the first part of this work. Indeed, one only needs to be able to numerically integrate the closed-loop dynamics of the system over a finite horizon under this backup law to compute all the information necessary for evaluating the regulation map and enforcing safety.
The effect of relaxing the stabilization requirements of the backup law is also studied, and weaker but more practical safety guarantees are brought forward. We then explore the relationship between the optimality of the backup law and how conservative the resulting safety filter is. Finally, methods of selecting a safe input with varying levels of trade-off between conservativeness and computational complexity are proposed and illustrated on multiple robotic systems, namely: a two-wheeled inverted pendulum (Segway), an industrial manipulator, a quadrotor, and a lower body exoskeleton.</p>https://resolver.caltech.edu/CaltechTHESIS:06022020-154234707Tethered Motion Planning for a Rappelling Robot
https://resolver.caltech.edu/CaltechTHESIS:06012020-230913819
Year: 2020
DOI: 10.7907/h7d4-ww72
<p>The Jet Propulsion Laboratory and Caltech developed the Axel rover to investigate and demonstrate the potential for tethered extreme terrain mobility, such as allowing access to science targets on the steep crater walls of other planets. Tether management is a key issue for Axel and other rappelling rovers. Avoiding tether entanglement constrains the robot's valid motions to the set of outgoing and returning path pairs that are homotopic to each other. In the case of a robot on a steep slope, a motion planner must additionally ensure that this ascent-descent path pair is feasible, based on the climbing forces provided by the tether. This feasibility check relies on the taut tether configuration, which is the shortest path in the homotopy class of the ascent-descent path pair. </p>
<p>This dissertation presents a novel algorithm for tethered motion planning in extreme terrains, produced by combining shortest-homotopic-path algorithms from the topology and computational geometry communities with traditional graph search methods. The resulting tethered motion planning algorithm searches for this shortest path, checks for feasibility, and then generates waypoints for an ascent-descent path pair in the same homotopy class. I demonstrate the implementation of this algorithm on a Martian crater data set such as might be seen for a typical mission. By searching only for the shortest path, and ordering that search according to a heuristic, this algorithm proceeds more efficiently than previous tethered path-planning algorithms for extreme terrain. </p>
<p>Frictional tether-terrain interaction may cause dangerously intermittent and unstable tether obstacles, which can be categorized based on their stability. Force-balance equations from the rope physics literature provide a set of tether and terrain conditions for static equilibrium, which can be used to determine if a given tether configuration will stick to a given surface based on tether tension. By estimating the tension of Axel's tether when driving, I divide potential tether tension obstacles into the following categories: acting as obstacles, acting as non-obstacles, and hazardous intermittent obstacles where it is uncertain whether the tether would slip or stick under normal driving tension variance. This dissertation describes how to modify the obstacle map as the categorization of obstacles fluctuates, and how to alter a motion plan around the dangerous tether friction obstacles. Together, these algorithms and methods form a framework for tethered motion planning on extreme terrain.</p>https://resolver.caltech.edu/CaltechTHESIS:06012020-230913819Scalable Synthesis and Verification: Towards Reliable Autonomy
https://resolver.caltech.edu/CaltechTHESIS:04292020-165136662
Year: 2020
DOI: 10.7907/4j39-v857
<p>We have seen the growing deployment of autonomous systems in our daily life, ranging from safety-critical self-driving cars to dialogue agents. While impactful and impressive, these systems do not often come with guarantees and are not rigorously evaluated for failure cases. This is in part due to the limited scalability of tools available for designing correct-by-construction systems, or verifying them posthoc. Another key limitation is the lack of availability of models for the complex environments with which autonomous systems often have to interact with. In the direction of overcoming these above mentioned bottlenecks to designing reliable autonomous systems, this thesis makes contributions along three fronts.</p>
<p>First, we develop an approach for parallelized synthesis from linear-time temporal logic Specifications corresponding to the generalized reactivity (1) fragment. We begin by identifying a special case corresponding to singleton liveness goals that allows for a decomposition of the synthesis problem, which facilitates parallelized synthesis. Based on the intuition from this special case, we propose a more generalized approach for parallelized synthesis that relies on identifying equicontrollable states.</p>
<p>Second, we consider learning-based approaches to enable verification at scale for complex systems, and for autonomous systems that interact with black-box environments. For the former, we propose a new abstraction refinement procedure based on machine learning to improve the performance of nonlinear constraint solving algorithms on large-scale problems. For the latter, we present a data-driven approach based on chance-constrained optimization that allows for a system to be evaluated for specification conformance without an accurate model of the environment. We demonstrate this approach on several tasks, including a lane-change scenario with real-world driving data.</p>
<p>Lastly, we consider the problem of interpreting and verifying learning-based components such as neural networks. We introduce a new method based on Craig's interpolants for computing compact symbolic abstractions of pre-images for neural networks. Our approach relies on iteratively computing approximations that provably overapproximate and underapproximate the pre-images at all layers. Further, building on existing work for training neural networks for verifiability in the classification setting, we propose extensions that allow us to generalize the approach to more general architectures and temporal specifications.</p>https://resolver.caltech.edu/CaltechTHESIS:04292020-165136662From Bipedal to Quadrupedal Locomotion, Experimental Realization of Lyapunov Approaches
https://resolver.caltech.edu/CaltechTHESIS:05042021-155258800
Year: 2021
DOI: 10.7907/j1ty-zb28
<p>Possibly one of the most significant innovations of the past decade is the hybrid zero dynamics (HZD) framework, which formally and rigorously designs a control algorithm for robotic walking. In this methodology, Lyapunov stability, which is often used to certificate a dynamical system's stability, was introduced to the control law design for a hybrid control system. However, the prerequisites of precise modeling to apply the HZD methodology can often be too restrictive to design controllers for uncertain and complex real-world hardware experiments. This thesis addresses the problem raised by noisy measurements and the intricate hybrid structure of locomotion dynamics.</p>
<p>First, the HZD methodology's construction is based on the full-order, hybrid dynamics of legged locomotion, which can be intractable for control synthesis for high-dimensional systems. This thesis studies the general structure of hybrid control systems for walking systems, ranging from 1D hopping, 2D walking, 2D running, and 3D quadrupedal locomotion on rough terrains. Further, we characterize a walking behavior--gait--as a solution (execution) to a hybrid control system. To find these solutions, which represent a "gait," we employed advanced numerical methods such as collocation methods to parse the solution-finding problem into the open- and closed-loop trajectory optimization problems. The result is that we can find versatile gaits for ten different robotic platforms efficiently. This includes bipedal running, bipedal walking on slippery surfaces, and quadrupedal robots walking on sloped terrains. The numerous solution-finding examples expand the applicability of the HZD framework towards more complex dynamical systems.</p>
<p>Further, for the uncertain and noisy real-world implementation, the exponential stability of the continuous dynamics is an ideal but restrictive condition for hybrid stability. This condition is especially challenging to satisfy for highly dynamical behaviors such as bipedal running, which loses ground support for a short period. This thesis observes the destabilizing effect of the noisy measurements of the phasing variable. By reformulating the traditional input-to-state stability (ISS) concept into phase-uncertainty to state stability, we are able to synthesize a robust controller for bipedal running on DURUS-2D. This time+state-based controller formally guarantees stability under noisy measurements and stabilizes the 1.75 m/s running experiments.</p>
<p>Lastly, robotic dynamics have long been characterized as the interconnection of rigid-body dynamics. We take this perspective one step further and incorporate controller design into the formulation of coupled control systems (CCS). We first view a quadrupedal robot as two bipedal robots connected via some holonomic constraints. In a dimensional reduction manner, we develop a novel optimization framework, and the computational performance is reduced to a few seconds for gait generation. Furthermore, we can design local controllers for each bipedal subsystem and still guarantee the overall system's stability. This is done by combining the HZD framework and the ISS properties to contain the disturbance induced by the other subsystems' inputs. Utilizing the proposed CCS methods, we will experimentally realize quadrupedal walking on various outdoor rough terrains.</p>https://resolver.caltech.edu/CaltechTHESIS:05042021-155258800Intelligent Control for Fixed-Wing eVTOL Aircraft
https://resolver.caltech.edu/CaltechTHESIS:02182021-040721884
Year: 2021
DOI: 10.7907/51c6-aa57
<p>Urban Air Mobility (UAM) holds promise for personal air transportation by deploying "flying cars" over cities. As such, fixed-wing electric vertical take-off and landing (eVTOL) aircraft has gained popularity as they can swiftly traverse cluttered areas, while also efficiently covering longer distances. These modes of operation call for an enhanced level of precision, safety, and intelligence for flight control. The hybrid nature of these aircraft poses a unique challenge that stems from complex aerodynamic interactions between wings, rotors, and the environment. Thus accurate estimation of external forces is indispensable for a high performance flight. However, traditional methods that stitch together different control schemes often fall short during hybrid flight modes. On the other hand, learning-based approaches circumvent modeling complexities, but they often lack theoretical guarantees for stability.</p>
<p>In the first part of this thesis, we study the theoretical benefits of these fixed-wing eVTOL aircraft, followed by the derivation of a novel unified control framework. It consists of nonlinear position and attitude controllers using forces and moments as inputs; and control allocation modules that determine desired attitudes and thruster signals. Next, we present a composite adaptation scheme for linear-in-parameter (LiP) dynamics models, which provides accurate realtime estimation for wing and rotor forces based on measurements from a three-dimensional airflow sensor. Then, we introduce a design method to optimize multirotor configuration that ensures a property of robustness against rotor failures.</p>
<p>In the second part of the thesis, we use deep neural networks (DNN) to learn part of unmodeled dynamics of the flight vehicles. Spectral normalization that regulates the Lipschitz constants of the neural network is applied for better generalization outside the training domain. The resultant network is utilized in a nonlinear feedback controller with a contraction mapping update, solving the nonaffine-in-control issue that arises. Next, we formulate general methods for designing and training DNN-based dynamics, controller, and observer. The general framework can theoretically handle any nonlinear dynamics with prior knowledge of its structure. Finally, we establish a delay compensation technique that transforms nominal controllers for an undelayed system into a sample-based predictive controller with numerical integration. The proposed method handles both first-order and transport delays in actuators and balances between numerical accuracy and computational efficiency to guarantee stability under strict hardware limitations.</p>https://resolver.caltech.edu/CaltechTHESIS:02182021-040721884Safe and Interpretable Autonomous Systems Design: Behavioral Contracts and Semantic-Based Perception
https://resolver.caltech.edu/CaltechTHESIS:04022021-033321217
Year: 2021
DOI: 10.7907/w3m8-es32
<p>We are on the verge of experiencing a new, integrated society where autonomous vehicles will become a fabric of our everyday lives. And yet, seamless integration of autonomous vehicles into our society will require vehicles to interface safely with humans in an incredibly complex, fast-paced, and dynamic environment. Premature deployment of these new autonomous systems — without safety guarantees or interpretability of algorithms, could prove catastrophic. How can algorithms governing vehicle behavior be designed in a way that guarantees safety, performance, interpretability and scalability? This is the question this thesis seeks to answer. </p>
<p>First, we present a framework for architecting the decision-making module of autonomous vehicles so that safety and progress of agents can be formally guaranteed. In particular, all agents are defined to act according to what is termed an assume-guarantee contract, which is broadly defined as a set of behavioral preferences. The first version of the assume-guarantee contract is a behavioral profile, which is a set of ordered rules that agents must use to select actions in a way that is interpretable. With all agents operating according to a behavioral profile, the interactions however, are not necessarily coordinated. We then constrain agent behavior with an additional set of interaction rules. The behavioral profile combined with these additional constraints, are what we term a behavioral protocol. With all agents operating according to a local, decentralized behavioral protocol, we can provide formal proofs of the correctness of agent behavior, i.e. all agents will never collide and agents will make it to their respective destinations. Not only does the protocol so defined allow us to make formal guarantees, but it is also designed in a way that scales well in the number of agents and provides interpretability of agent behaviors. Safety and progress guarantees are proven and verified in simulation. </p>
<p>Second, we focus on using information from object classifiers to enhance an autonomous vehicle's ability to localize where it is within its environment. The proposed approach for incorporating this semantic information is based on solving the maximum likelihood problem. With a hierarchical formulation, we are not only able to improve upon the accuracy of traditional localization techniques, but we are also able to improve our confidence in the accuracy of object detection classifications. The improvement in robustness and accuracy of these algorithms are shown in simulation.</p>https://resolver.caltech.edu/CaltechTHESIS:04022021-033321217Reduced Order Model Inspired Robotic Bipedal Walking: A Step-to-step Dynamics Approximation based Approach
https://resolver.caltech.edu/CaltechTHESIS:06022021-035141903
Year: 2021
DOI: 10.7907/9bz9-x102
<p>Controlling bipedal robotic walking is a challenging task. The dynamics is hybrid, nonlinear, high-dimensional, and typically underactuated. Complex physical constraints have to be satisfied in the walking generation. The stability in terms of not-falling is also hard to be encoded in the walking synthesis. Canonical approaches for enabling robotic walking typically rely on large-scale trajectory optimizations for generating optimal periodic behaviors on the full-dimensional model of the system; then the stabilities of the controlled behaviors are analyzed through the numerically derived Poincaré maps. This full-dimensional periodic behavior based synthesis, despite being theoretically rigorous, suffers from several disadvantages. The trajectory optimization problem is computationally challenging to solve. Non-trivial expert-tuning is required on the cost, constraints, and initial conditions to avoid infeasibilities and local optimality. It is cumbersome for realizing non-periodical behaviors, and the synthesized walking can be sensitive to model uncertainties.</p>
<p>In this thesis, we propose an alternative approach of walking synthesis that is based on reduced order modeling and dynamics approximation. We formulate a discrete step-to-step (S2S) dynamics of walking, where the step size is treated as the control input to stabilize the pre-impact horizontal center of mass (COM) state of the robot. Stepping planning thus is converted into a feedback control problem. To effectively and efficiently solve this feedback stepping planning problem, an underactuated Hybrid Linear Inverted Pendulum (H-LIP) model is proposed to approximate the dynamics of underactuated bipedal walking; the linear S2S dynamics of the H-LIP then approximates the robot S2S dynamics. The H-LIP based stepping controller is hence utilized to plan the desired step sizes on the robot to control its pre-impact horizontal COM state. Stable walking behaviors are consequently generating by realizing the desired step size in the output construction and stabilizing the output via optimization-based controllers. We evaluate this approach successfully on several bipedal walking systems with an increase in the system complexity: a planar five-linkage walker AMBER, an actuated version of the Spring Loaded Inverted Pendulum (aSLIP) in both 2D and 3D, and finally the 3D underactuated robot Cassie. The generated dynamic walking behaviors on these systems are shown to be highly versatile and robust. Furthermore, we show that this approach can be effectively extended to realizing more complex walking tasks such as global trajectory tracking and walking on rough terrain.</p>https://resolver.caltech.edu/CaltechTHESIS:06022021-035141903Bioinspired Nanostructures for Biomedical Applications
https://resolver.caltech.edu/CaltechTHESIS:07242020-111050846
Year: 2021
DOI: 10.7907/atnt-8p46
<p>Nature boasts a myriad examples of coloration achieved purely through the physical interaction of light with nano-scale features also known as biophotonic nanostructures. From reptiles to insects, birds to flora, structural coloration has been achieved through a variety of fascinating nano-architectures that leverage different physics. Beyond structural coloration, these nanostructures are often truly multifunctional. For instance, biophotonic nanostructures can also serve as self-cleaning and bactericidal surfaces, gas and thermal sensors, waveguides and beam splitters. With the growing need for robust and compact biomedical devices, the requirement to embed multiple functionalities towards sensing, monitoring, diagnostics and therapeutics within a diminutive device footprint becomes crucial. In this regard, inspiration from the multifunctionality of biophotonic nanostructures can prove to be greatly beneficial for medical applications. Consequently, this work attempts to showcase various examples of the utilization of nanostructures inspired from biophotonic nanostructures for biomedical applications under various overlapping themes such as ophthalmic sensors, bioinspired optics and plasmonic biosensing.</p>
<p>This thesis is summarized in two parts. The first part (Chapters 2--4) introduces a proof-of-concept optical intraocular pressure (IOP) sensor implant and various challenges faced during its <i>in vivo</i> implementation. In Chapter 3, nanostructures inspired by light-trapping epidermal micro-/nanostructures on flower petals are proposed and embedded onto the sensor platform to improve its <i>in vivo</i> optical signal-to-noise ratio and biocompatibility. Chapter 4 covers nanostructures inspired by biophotonic nanostructures on longtail glasswing butterfly wings that improve the <i>in vivo</i> angle of acceptance and biocompatibility of the sensor.</p>
<p>The second part (Chapters 5 and 6) presents the use of bioinspired nanostructures in plasmonic biosensors. Chapter 5 discusses an on-chip platform consisting of bioinspired plasmonic nanostructures to detect various nucleic acid sequences of relevance in the pathogenesis of HIV-1 via plasmon-enhanced fluorescence. Chapter 6 describes the employment of bioinspired quasi-ordered nanostructuring on flexible substrates for broadband surface-enhanced Raman spectroscopy (SERS). Here, SERS-based biosensing enabled by quasi-ordering is used to detect uric acid -- a biomarker of various pathologies in human tears.</p>https://resolver.caltech.edu/CaltechTHESIS:07242020-111050846Assuring Safety under Uncertainty in Learning-Based Control Systems
https://resolver.caltech.edu/CaltechTHESIS:01052021-195655093
Year: 2021
DOI: 10.7907/9kye-rn93
<p>Learning-based controllers have recently shown impressive results for different robotic tasks in well-defined environments, successfully solving a Rubiks cube and sorting objects in a bin. These advancements promise to enable a host of new capabilities for complex robotic systems. However, these learning-based controllers cannot yet be deployed in highly uncertain environments due to significant issues relating to learning reliability, robustness, and safety.</p>
<p>To overcome these issues, this thesis proposes new methods for integrating model information (e.g. model-based control priors) into the reinforcement learning framework, which is crucial to ensuring reliability and safety. I show, both empirically and theoretically, that this model information greatly reduces variance in learning and can effectively constrain the policy search space, thus enabling significant improvements in sample complexity for the underlying RL algorithms. Furthermore, by leveraging control barrier functions and Gaussian process uncertainty models, I show how system safety can be maintained under uncertainty without interfering with the learning process (e.g. distorting the policy gradients).</p>
<p>The last part of the thesis will discuss fundamental limitations that arise when utilizing machine learning to derive safety guarantees. In particular, I show that widely used uncertainty models can be highly inaccurate when predicting rare events, and examine the implications of this for safe learning. To overcome some of these limitations, a novel framework is developed based on assume-guarantee contracts in order to ensure safety in multi-agent human environments. The proposed approach utilizes contracts to impose loose responsibilities on agents in the environment, which are learned from data. Imposing these responsibilities on agents, rather than treating their uncertainty as a purely random process, allows us to achieve both safety and efficiency in interactions.</p>https://resolver.caltech.edu/CaltechTHESIS:01052021-195655093Contract-Based Design: Theories and Applications
https://resolver.caltech.edu/CaltechTHESIS:01132021-065636010
Year: 2021
DOI: 10.7907/8vp7-kd82
<p>Most things we know only exist in relation to one another. Their states are strongly coupled due to dependencies that arise from such relations. For a system designer, acknowledging the presence of these dependencies is as crucial to guaranteeing performance as studying them. As the roles played by technology in fields such as transportation, healthcare, and finance continue to be more profound and diverse, modern engineering systems have grown to be more reliant on the integration of technologies across multiple disciplines and their requirements. The need to ensure proper division of labor, integration of system modules, and attribution of legal responsibility calls for a more methodological look into co-design considerations. Originally conceived in computer programming, contract-based reasoning is a design approach whose promise of a formal compositional paradigm is receiving attention from a broader engineering community. Our work is dedicated to narrowing the gap between the theory and application of this yet nascent framework.</p>
<p>In the first half of this dissertation, we introduce a model interface contract theory for input/output automata with guards and a formalization of the directive-response architecture using assume-guarantee contracts and show how these may be used to guide the formal design of a traffic intersection and an automated valet parking system respectively. Next, we address a major drawback of assume-guarantee contracts, i.e., the problem of a void contract due to antecedent failure. Our proposed solution is a reactive version of assume-guarantee contracts that enables direct specification at the assumption and guarantee level along with a novel synthesis algorithm that exposes the effects of failures on the contract structure. This is then used to help optimize, adapt, and robustify our design against an uncertain environment.</p>
<p>In light of ongoing development of autonomous driving technologies and its potential impact on the safety of future transportation, the second half of this work is dedicated to the application of the design-by-contract framework to the distributed control of autonomous vehicles. We start by defining and proving properties of "assume-guarantee profiles," our proposed approach to transparent distributed multi-agent decision making and behavior prediction. Next, we provide a local conflict resolution algorithm in the context of a quasi-simultaneous game which guarantees safety and liveness to the composition of autonomous vehicle systems in this game. Finally, to facilitate the extension of these frameworks to real-life urban driving settings, we also supply an effective method to predict agent behavior that utilizes recent advances in machine learning research.</p>https://resolver.caltech.edu/CaltechTHESIS:01132021-065636010Towards Learning Robotic Dynamics: Application to Multirotor Takeoff and Landing
https://resolver.caltech.edu/CaltechTHESIS:03152021-082447788
Year: 2021
DOI: 10.7907/199j-dk87
<p>Multirotors have become widespread but their usage is still limited. Ensuring safety during take-off and landing is still an open problem. Towards this goal this thesis proposes two different solutions to address this problem. The two approaches complement each other and they are tested on hardware.</p>
<p>The first approach is to design a vehicle that is stable during take-off, despite hardware failures or unsteady take-off platforms. A solution is to use a ballistic launch to impose a deterministic path, preventing collisions with its environment. Following this approach led to the development of several SQUID (<i>Streamlined Quick Unfolding Investigation Drone</i>) vehicles. The main challenges are the ballistic initial flight, large accelerations during launch, and limited volume. A first prototype was developed, which is able to transition mid-flight from stable ballistic flight to a fully controllable multirotor. The system has been fabricated and field tested from a moving vehicle up to 50mph to successfully demonstrate the feasibility of the concept and experimentally validate the design's aerodynamic stability and deployment reliability. A second prototype expanded the first one's capabilities incorporating fully-autonomous vision-based navigation, while keeping the ballistic passive stability and stable transition abilities. The new design includes a more reliable plate-based structure and more effective folding fins.</p>
<p>The second approach focuses on designing controllers that are safe regardless of the platform. For that purpose, a Model Predictive Control (MPC) is used to ensure state and input constraints. Given the highly non-linear dynamics platforms and fast dynamics that require a quick controller evaluation, the work in this thesis is built using Koopman Operator theory, which allows tools from linear analysis to be applied to systems with inherently non-linear dynamics. One of the main contributions is a novel method to find Koopman Eigenfunctions directly from data. Another key contribution is an episodic approach to model non-linear actuation dynamics. The proposed method is first tested on simulation and it outperforms comparable approaches. The method is also demonstrated on-board a multirotor for a fast landing application, where the nonlinear ground effect is learned and used to improve landing speed and quality. An additional extension considers model uncertainty in the MPC architecture, where an Ensemble Kalman Sampler is used to learn the uncertainty distribution.</p>https://resolver.caltech.edu/CaltechTHESIS:03152021-082447788Dynamic Bipedal Locomotion: From Hybrid Zero Dynamics to Control Lyapunov Functions via Experimentally Realizable Methods
https://resolver.caltech.edu/CaltechTHESIS:05282021-062435188
Year: 2021
DOI: 10.7907/h8v0-vd47
<p>Robotic bipedal locomotion has become a rapidly growing field of research as humans increasingly look to augment their natural environments with intelligent machines. In order for these robotic systems to navigate the often unstructured environments of the world and perform tasks, they must first have the capability to dynamically, reliably, and efficiently locomote. Due to the inherently hybrid and underactuated nature of dynamic bipedal walking, the greatest experimental successes in the field have often been achieved by considering all aspects of the problem; with explicit consideration of the interplay between modeling, trajectory planning, and feedback control.</p>
<p>The methodology and developments presented in this thesis begin with the modeling and design of dynamic walking gaits on bipedal robots through hybrid zero dynamics (HZD), a mathematical framework that utilizes hybrid system models coupled with nonlinear controllers that results in stable locomotion. This will form the first half of the thesis, and will be used to develop a solid foundation of HZD trajectory optimization tools and algorithms for efficient synthesis of accurate hybrid motion plans for locomotion on two underactuated and compliant 3D bipeds. While HZD and the associated trajectory optimization are an existing framework, the resulting behaviors shown in these preliminary experiments will extend the limits of what HZD has demonstrated is possible thus far in the literature. Specifically, the core results of this thesis demonstrate the first experimental multi-contact humanoid walking with HZD on the DURUS robot and then through the first compliant HZD motion library for walking over a continuum of walking speeds on the Cassie robot.</p>
<p>On the theoretical front, a novel formulation of an optimization-based control framework is introduced that couples convergence constraints from control Lyapunov functions (CLF)s with desirable formulations existing in other areas of the bipedal locomotion field that have proven successful in practice, such as inverse dynamics control and quadratic programming approaches. The theoretical analysis and experimental validation of this controller thus forms the second half of this thesis. First, a theoretical analysis is developed which demonstrates several useful properties of the approach for tuning and implementation, and the stability of the controller for HZD locomotion is proven. This is then extended to a relaxed version of the CLF controller, which removes a convergence inequality constraint in lieu of a conservative CLF cost within a quadratic program to achieve tracking. It is then explored how this new CLF formulation can fully leverage the planned HZD walking gaits to achieve the target performance on physical hardware. Towards this goal, an experimental implementation of the CLF controller is derived for the Cassie robot, with the resulting experiments demonstrating the first successful realization of a CLF controller for a 3D biped on hardware in the literature. The accuracy of the robot model and synthesized HZD motion library allow the real-time control implementation to regularize the CLF optimization cost about the nominal walking gait. This drives the controller to choose smooth input torques and anticipated spring torques, as well as regulate an optimal distribution of feasible ground reaction forces on hardware while reliably tracking the planned virtual constraints. These final results demonstrate how each component of this thesis were brought together to form an effective end-to-end implementation of a nonlinear control framework for underactuated locomotion on a bipedal robot through modeling, trajectory optimization, and then ultimately real-time control.</p>https://resolver.caltech.edu/CaltechTHESIS:05282021-062435188Online Learning from Human Feedback with Applications to Exoskeleton Gait Optimization
https://resolver.caltech.edu/CaltechTHESIS:12092020-162149429
Year: 2021
DOI: 10.7907/gvtx-1586
<p>Systems that intelligently interact with humans could improve people's lives in numerous ways and in numerous settings, such as households, hospitals, and workplaces. Yet, developing algorithms that reliably and efficiently personalize their interactions with people in real-world environments remains challenging. In particular, one major difficulty lies in adapting to human-in-the-loop feedback, in which an algorithm makes sequential decisions while receiving online feedback from humans; throughout this interaction, the algorithm seeks to optimize its decision-making quality, as measured by the utility of its performance to the human users. Such algorithms must balance between exploration and exploitation: on one hand, the algorithm must select uncertain strategies to fully explore the environment and the interacting human's preferences, while on the other hand, it must exploit the empirically-best-performing strategies to maximize its cumulative performance.</p>
<p>Learning from human feedback can be difficult, as people are often unreliable in specifying numerical scores. In contrast, humans can often more accurately provide various types of qualitative feedback, for instance pairwise preferences. Yet, sample efficiency is a significant concern in human-in-the-loop settings, as qualitative feedback is less informative than absolute metrics, and algorithms can typically pose only limited queries to human users. Thus, there is a need to create theoretically-grounded online learning algorithms that efficiently, reliably, and robustly optimize their interactions with humans while learning from online qualitative feedback.</p>
<p>This dissertation makes several contributions to algorithm design for human-in-the-loop learning. Firstly, this work develops the Dueling Posterior Sampling (DPS) algorithmic framework, a model-based, Bayesian approach for online learning in the settings of preference-based reinforcement learning and generalized linear dueling bandits. DPS is developed together with a theoretical regret analysis framework, and yields competitive empirical performance in a range of simulations. Additionally, this thesis presents the CoSpar and LineCoSpar algorithms for sample-efficient, mixed-initiative learning from pairwise preferences and coactive feedback. CoSpar and LineCoSpar are both deployed in human subject experiments with a lower-body exoskeleton to identify optimal, user-preferred exoskeleton walking gaits. This work presents the first demonstration of preference-based learning for optimizing dynamic crutchless exoskeleton walking for user comfort, and makes progress toward customizing exoskeletons and other assistive devices for individual users.</p>https://resolver.caltech.edu/CaltechTHESIS:12092020-162149429Spacecraft Motion Planning and Control under Probabilistic Uncertainty for Coordinated Inspection and Safe Learning
https://resolver.caltech.edu/CaltechTHESIS:05142021-163257155
Year: 2021
DOI: 10.7907/6329-sf68
<p>During a spacecraft mission design process, engineers often balance the following three criteria: science return, optimality in performance, and safety. Given a science criterion, engineers design the orbit parameters with predefined performance and safety. Often in this approach, the spacecraft has no understanding of the expected outcome or the knowledge of the mission safety criteria. Autonomous science-driven orbit (or goal) selection and planning for safety under uncertainty enable efficient and adaptable missions. To this end, we propose an architecture for information-based guidance and control for coordinated inspection, motion planning and control algorithms for safe and optimal guidance under uncertainty, and architecture for safe exploration.</p>
<p>In the first part of this thesis, we present an architecture for inspection or mapping of a target spacecraft in a low Earth orbit using multiple observer spacecraft. We use an information gain approach to directly consider the trade-off between gathered data and fuel/energy cost. The estimated information gain is a crucial input to the motion planner, which computes orbits and reconfiguration strategies for each of the observers to maximize the information gain from distributed observations of the target spacecraft. The resulting motion trajectories jointly consider observational coverage of the target spacecraft and fuel/energy cost. We validate our architecture in a mission simulation to visually inspect the target spacecraft and on the three degree-of-freedom robotic spacecraft dynamics simulator testbed.</p>
<p>In the second part of the thesis, we present gPC-SCP, Generalized Polynomial Chaos-based Sequential Convex Programming method, to compute a sub-optimal solution for a continuous-time chance-constrained stochastic nonlinear optimal control (SNOC) problem. The approach enables motion planning and control of robotic systems under uncertainty. The proposed method involves two steps. The first step is to derive a deterministic nonlinear optimal control problem (DNOC) with convex constraints that are surrogate to the SNOC by using gPC expansion and the distributionally-robust convex subset of the chance constraints. The second step is to solve the DNOC problem using sequential convex programming (SCP) for trajectory generation and control. We prove that in the unconstrained case, the optimal value of the DNOC converges to that of SNOC asymptotically and that any feasible solution of the constrained DNOC is a feasible solution of the chance-constrained SNOC. We derive a stable stochastic model predictive controller using the gPC-SCP for tracking a potentially unsafe trajectory in the presence of uncertainty. We empirically demonstrate the efficacy of the gPC-SCP method for the following three test cases: 1) collision checking under uncertainty in actuation, 2) collision checking with stochastic obstacles, and 3) safe trajectory tracking under uncertainty in the dynamics and obstacle location by using a receding horizon control approach. We validate the effectiveness of the gPC-SCP method on the robotic spacecraft testbed.</p>
<p>In the third part of this thesis, we present a new approach for optimal motion planning for safe exploration that integrates the chance-constrained stochastic optimal control with dynamics learning and feedback control. We derive an iterative convex optimization algorithm that solves an Information-cost Stochastic Nonlinear Optimal Control problem (Info-SNOC). The optimization objective encodes control cost for performance and exploration cost for learning, and the safety is incorporated as distributionally robust chance constraints. The dynamics are predicted from a robust regression model that is learned from data. The Info-SNOC algorithm is used to compute a sub-optimal pool of safe motion plans that aid in exploration for learning unknown residual dynamics under safety constraints. A stable feedback controller is used to execute the motion plan and collect data for model learning. We prove the safety of rollout from our exploration method and reduction in uncertainty over epochs, thereby guaranteeing the consistency of our learning method. We validate the effectiveness of Info-SNOC by designing and implementing a pool of safe trajectories for a planar robot. We demonstrate that our approach has a higher success rate in ensuring safety when compared to a deterministic trajectory optimization approach.</p>https://resolver.caltech.edu/CaltechTHESIS:05142021-163257155Bidirectional Brain-Machine Interfaces for Modulating Stimulation and Neural Plasticity
https://resolver.caltech.edu/CaltechTHESIS:08272021-174548241
Year: 2022
DOI: 10.7907/df0t-1t79
<p>In prosthetics, tactile feedback can let us feel how we interact with the environment. Without this, it is extremely difficult to perform a motor task with fine control. The same idea can be applied in the brain-machine interface (BMI), which is an interface that directly connects external devices such as prosthetic limbs to the brain. Bidirectional BMI can deliver a stimulation to the brain as a sensory feedback, which can improve the performance of motor tasks. Such a bidirectional BMI can also serve a different role, if the stimulation encodes different information: if it encodes neural activity from another brain area, for example, then bidirectional BMI can provide a bypass for a damaged neural circuit. This may also affect the neural connectivity, strengthening or weakening the underlying neural connections. In this thesis, we present experiments that explore such applications of bidirectional BMI. First, we describe an experiment for characterizing neural connectivity between different brain areas. We found neural connectivity between supramarginal gyrus (SMG) and PMv (ventral premotor area), and also between anterior intraparietal (AIP) and Brodmann’s area 5 (BA5), characterized by field-field, spike-field, and partial spike-field coherence. Through partial spike-field coherence, we also revealed that the spikes in PMv may drive the activity in SMG, which is obscured in ordinary spike-field coherence. Next, we provide evidence of changes in neural connectivity caused by stimulation in S1. With spike-triggered stimulation, which delivers stimulation in S1 in response to spikes recorded in a selected channel in SMG, we could significantly increase the correlation between SMG and S1, measured by the spike time tilling coefficient (STTC) to avoid dependencies of the correlation on firing rates. Furthermore, we found that not only spike-triggered stimulations, but also random stimulations on multiple channels in S1, can vary partial spike-field coherence in theta and alpha bands within S1; such changes mostly occurred in channel pairs with zero phase difference in partial spike-field coherence. Finally, we demonstrate the possibility of volitional control on stimulation pattern in bidirectional BMI. It is shown that the participants could not only increase or decrease a single-channel firing rate, but also hold the firing rate in a given range, demonstrating a fine control over firing rate. These findings would begin to establish a framework for closed-loop modulation of neural activity with bidirectional BMI and could be used to develop new treatments for neurological damage, such as to promote plasticity in or bridge brain areas affected by stroke.</p>https://resolver.caltech.edu/CaltechTHESIS:08272021-174548241Koopman-based Learning and Control of Agile Robotic Systems
https://resolver.caltech.edu/CaltechTHESIS:10122021-213903517
Year: 2022
DOI: 10.7907/2t6d-j206
<p>Learning methods to enable high performance control systems have recently shown promising results in selected environments and applications. These advances promote the next generation of autonomous robots capable of significantly improving efficiency, cost, and safety in their respective domains. Importantly, these systems are <i>safety-critical</i> and operate in proximity to humans in diverse and uncertain environments. As a result, operational failures may cause significant material and societal losses. Additionally, robot learning and control are further complicated by requiring fast controller update rates and operational constraint satisfaction.</p>
<p>To address these challenges, this thesis presents multiple methods based on Koopman operator theory. The first approach develops algorithms to learn lifted-dimensional models of nonlinear systems and leverages the models in model predictive control (MPC) design. Koopman-based methods typically employ hand-crafted observable functions to "lift" the state variables to the higher dimensional space. For most systems, this leads to poor prediction performance and inefficient use of data and computational resources. Instead, I present methods that generate observable functions from data, both based on underlying theory and by incorporating the observable functions and model structure in a neural network model. This allows lower dimensional models, important for real-time control, and enables the nonlinearities of control-affine dynamics to be captured, crucial to describing many robotic systems. I use quadrotor drones to experimentally demonstrate that the learned models combined with MPC can achieve close to optimal behavior while respecting important operational constraints.</p>
<p>The last part of the thesis is concerned with endowing systems with an arbitrary nominal control policy with safety guarantees. Control barrier functions (CBFs) are a powerful tool to achieve this, yet they rely on the computation of control invariant sets, which is notoriously difficult. To avoid this, a backup strategy can be used to implicitly define a control invariant set. However, this requires forward integration of the system dynamics under a backup controller, which is prohibitively expensive for realistic systems. I present a method that replaces the expensive integration using learned Koopman operators of the closed-loop dynamics. As a result, the online computation time required to evaluate the controller is drastically reduced, enabling real-time use. I also derive an error bound on the unmodeled dynamics in order to robustify the CBF controller and demonstrate the method on multi-agent collision avoidance for wheeled robots and quadrotors.</p>https://resolver.caltech.edu/CaltechTHESIS:10122021-213903517Felt, Imagined, and Seen Touch Share a Substrate in Human Posterior Parietal Cortex
https://resolver.caltech.edu/CaltechTHESIS:05022022-175609842
Year: 2022
DOI: 10.7907/0tgv-t705
One of the most remarkable aspects of human cognition is its flexibility. We can think new thoughts, infer meaning, plan actions, predict, extrapolate, and so much more. How do our brains enable this versatility? A growing ability to simultaneously record from large populations of single neurons in human cortex has begun to provide insight. Recent studies have identified that shared populations of neurons in posterior parietal cortex (PPC) of a human subject (involved in a brain-machine interface (BMI) clinical trial) encode many aspects of motor cognition: attempted and imagined actions, observed actions and the semantic processing of action verbs. Individual units are complex, but population representations manifest rich associations across neurons, supporting diverse behavioral contexts. Here, in novel work, we establish that the same PPC substrate also encodes aspects of sensory cognition, and unpack the functional organization of information that enables this versatility. We record populations of neurons in PPC of the same human subject, a tetraplegic trial participant implanted with a 4x4 mm microelectrode array. In a series of novel results, we first establish that neurons in this PPC substrate encode actual (or felt) touch to oneself, at short latency, with bilateral receptive fields, organized by body-part. We show that imagined touch to oneself and observed touch to others engage the same substrate. To understand coding mechanisms further, we manipulated the touch location (cheek, shoulder), and the touch type (pinch, press, rub, tap). As in the motor domain, individual neurons exhibit highly variable responses. At the population-level, however, we find that the diverse touch conditions are explained by a small number of subspaces (meaningful groupings of neurons) that encode basic-level, elemental information such as touch location, and touch type. This suggests a compositional basis in PPC, such that various touch conditions are encoded through diverse combinations of common primitive elements. Moreover, these subspaces are generalizable, able to explain novel (held out) data. These principles of compositionality and generalizability suggest a basis by which PPC may support cognitive behaviors such as comprehension, in situations that extend beyond our experiences. In support of this interpretation, we show finally that this PPC substrate encodes seen touch universally – not only to insensate arm regions on the tetraplegic human subject, and to other human individuals, but also to a wide sampling of inanimate objects. As predicted, neural information combines and generalizes across conditions such that touch to objects with more similar features, is more similarly encoded. Taken together, our work is a novel, neuron-level characterization of how high-level cortex in humans may support diverse sensory, motor, and cognitive behaviors. We speculate that populations of neurons in PPC encode rich internal models of the world that can be flexibly repurposed for diverse (and novel) behavioral contexts.https://resolver.caltech.edu/CaltechTHESIS:05022022-175609842Autonomous Mission-Driven Robots in Extreme Environments
https://resolver.caltech.edu/CaltechTHESIS:05172022-043237609
Year: 2022
DOI: 10.7907/a78d-kv42
<p>Robotic autonomy systems that can negotiate harsh environments under time and communication constraints are critical to accomplishing many real-world missions. Such systems require an integrated software-hardware solution capable of robustly reasoning about a time-limited mission across a complex environment and negotiating extreme physical conditions during mission execution. To this end, I will discus the development of two field-tested systems designed for operation in GPS-denied areas: (i) a coverage planning framework that enables efficient exploration of large, unknown environments, and (ii) a ballistically-launched aircraft that converts to an autonomous, free-flying multirotor in order to provide rapid aerial surveillance.</p>
<p>The first system addresses the time-limited exploration problem by providing a planning strategy that seeks to maximize the area covered by a robot’s sensor footprint along a planned trajectory. In order to find solutions over large spatial extents (>1 km) and long temporal horizons (>1 hour), this coverage problem is decomposed into tractable subproblems by introducing spatial and temporal abstractions. Spatially, the robot-world belief is approximated by a task-dependent structure, enriched with environment map estimates. Temporally, the belief is approximated by the aggregation of multiple structures, each spanning a different spatial range. Cascaded uncertainty-aware solvers return a coverage plan over the stratified belief in real time.
Coverage policies are constructed in a receding horizon fashion to ensure motion smoothness and resiliency to real-world stochasticity in perception and control. This coverage planning framework was extensively tested on physical robots in various real-world environments (caves, mines, subway systems, etc.) and served as the exploration strategy for a competing entry in the DARPA Subterranean Challenge.</p>
<p>The second system addresses rapid multirotor deployment for aerial data collection during emergencies. While multirotors are advantageous over fixed-winged systems due to their high maneuverability, their rotating blades are hazardous and require stable, uncluttered takeoff sites. To overcome this issue, a ballistically-launched, autonomously-stabilizing multirotor (SQUID -- Streamlined Quick Unfolding Investigation Drone) was designed, fabricated, and tested. SQUID follows a deterministic trajectory, transitioning from a folded launch configuration to an autonomous, fully-controllable hexacopter. The entire process from launch to position stabilization requires no user- or GPS-input and demonstrates the viability of using ballistically-launched multirotors to achieve safe and rapid deployment from moving vehicles.</p>https://resolver.caltech.edu/CaltechTHESIS:05172022-043237609Creating ARCHER: A 3D Hopping Robot with Flywheels for Attitude Control
https://resolver.caltech.edu/CaltechTHESIS:06012022-061623441
Year: 2022
DOI: 10.7907/gbts-va63
<p>The field of robotic hopping began over 40 years ago, when it was first shown that robust hopping could be achieved on real hardware. In the years since then, it's become clear that hopping requires high performance and precision from its actuation and planning, due to its extreme interactions with the environment occurring over periodic, yet very short durations of time. Despite being of lower dimensionality than many other legged robots, hoppers are very underactuated, which only adds to the difficulty of planning motions quickly for real-time needs.</p>
<p>The studies of robotic hopping presented in this thesis start with a look into two different actuation styles for creating vertical periodic motion: a compress-release mechanism and a moving-mass mechanism. The dynamics of each were examined from the perspective of stability and robustness to uncertainties in the model and measurements. The compress-release hopper (CRH) was found to be very stable, simple to control, and robust to all uncertainties, but inherently had some inefficiencies due to the requirement of holding compression during portions of the aerial phase. The moving-mass hopper (MMH) required optimization to generate the proper cyclic motions as well as closed-loop control to make them stable. Furthermore, the original configuration of the MMH was also less energetically efficient and robust to uncertainty than the CRH.</p>
<p>In an effort to improve the efficiency of the MMH, a second-generation robot was designed using the principle of parallel elasticity. This involved placing a second spring in parallel to the actuator which would naturally guide the motion of the moving-mass into an optimal path, eliminating a significant portion of actuation effort and improving the overall efficiency. An added benefit of this change was that the robot no longer required closed loop control to create stable hopping. This new robot was built and tested in the lab showing a dramatic improvement over the previous design. The principle of controlling the compliance in the actuator for efficient motion was then taken one step further by creating custom, nonlinear stiffness springs which would provide a more ideal trajectory of motion. This process utilized a design-in-the-loop optimization strategy that would both design these springs as well as the motions of the moving-mass to yield better actuation efficiency. A set of these springs was created and attached to the second-gen MMH, replacing the lower spring, and tested in the lab. These springs did slightly improve the efficiency of the robot, but were restricted by the material selection of the springs due to manufacturing limitations.</p>
<p>Moving into the realm of 3-Dimensional hopping, a final robot was designed and built: ARCHER. Unlike traditional hopping robots which use a torso with very large inertia to control the leg motion and balance, ARCHER uses a set of three flywheels. The goal of this robot was twofold: to study the feasibility of using flywheels alone to control attitude, and to take advantage of the principle of decoupled systems. By using strategically placed flywheels, the dynamics of the leg and the attitude subsystems were decoupled, meaning their actuation did not have a direct influence on each other. This allows for simpler motion planning and control. The culmination of this thesis was running experiments with this robot, showing its initial performance and ability to hop with separate controllers for each subsystem.</p>https://resolver.caltech.edu/CaltechTHESIS:06012022-061623441Autonomous Temporal Understanding and State Estimation during Robot-Assisted Surgery
https://resolver.caltech.edu/CaltechTHESIS:05272022-171138586
Year: 2022
DOI: 10.7907/n58k-tr61
<p>Robot-Assisted Surgery (RAS) has become increasingly important in modern surgical practice for its many benefits and advantages for both the patient and the healthcare professionals, as compared to traditional open surgeries and minimally invasive surgeries such as laparoscopy. Artificial intelligence applications during RAS and post-operative analysis can provide various surgeon-assisting functionalities and could potentially achieve a better surgery outcome. These applications, ranging from providing surgeons with advisory information during RAS and post-operative analysis to virtual fixture and supervised autonomous surgical tasks, share a necessary prerequisite of a comprehensive understanding of the current surgical scene. This understanding should include the knowledge of the current surgical task being performed, the surgeon's actions and gestures, the state of the patient, etc. Currently, there is yet to be a unified effort to achieve the autonomous temporal understanding and perception of an RAS at the high accuracy and efficiency required in the highly safety-critical field of medicine.</p>
<p>This thesis develops novel modeling methodologies and deep learning-based models for the autonomous perception and temporal segmentation of the current surgical scene during an RAS. An RAS procedure is modeled as a hierarchical system consisting of discrete surgical states at multiple levels of temporal granularity. These surgical states take the form of surgical tasks, operational steps, fine-grained surgical actions, etc. A broad range of computational experiments were performed to develop methods that achieve an accurate, robust, and efficient estimation of these surgical states. Multiple novel deep learning-based models for feature extraction, noise elimination, and efficient training were proposed and tested. This thesis also shows the significant benefits of incorporating multiple types of data streams recorded by the surgical robotic system to a more accurate surgical state estimation effort.</p>
<p>Two new RAS datasets that contains real-world RAS procedures and diverse experimental settings were collected and annotated--filling a gap in the data sets available for the development and testing of of robust surgical state estimation models. The performance and robustness of models in this thesis work were showcased with these highly complex and dynamic real-world RAS datasets and compared against state-of-the-art methods. A significant model performance improvement was observed in both surgical state estimation accuracy and efficiency. The modeling methodologies and deep learning-based models developed in this work have diverse potential applications to the development of a next-generation surgical robotic systems.</p>https://resolver.caltech.edu/CaltechTHESIS:05272022-171138586Safe Input Regulation for Robotic Systems
https://resolver.caltech.edu/CaltechTHESIS:06022022-064735213
Year: 2022
DOI: 10.7907/zz10-gv06
<p>The safety of robotic systems is paramount to their continued emergence into our lives. From collaborative industrial manipulators to drone deliveries to autonomous vehicles, safety is the primary concern when it comes to the continued adoption of these technologies. While a number of techniques can be used to design safe controllers and planners that govern the actions of these robots, few are able to provide the type of safety guarantee needed to bring these technologies into reality.</p>
<p>The goal of this thesis is to provide a framework for regulating, or filtering, existing control inputs before they are applied by the robot, in order to ensure that safety is upheld. To illustrate this, consider one of the primary applications for this method: human-operated robotic platforms. For vehicles, this framework would modify the throttle, braking, and steering commands from a human driver to prevent him from driving off the road or into other cars. However, when the human is operating the vehicle safely, his commands should go unaltered. This illustrates the idea of a minimally invasive safety regulator: one that only engages when absolutely necessary to ensure safety.</p>
<p>Within the last decade, the mathematical framework that allows us to achieve this result, control barrier functions, was introduced. Its adoption among the nonlinear controls community has been rapid, and the method has been used to create controllers that guarantee safety on a large class of systems. Despite this, real-world implementations of control barrier functions are less common, since they require a very accurate model of the system, and they can be difficult to formulate properly. This work provides several major extensions, improvements, and modifications of control barrier functions that allow them to be utilized on a variety of real-world robotic systems.</p>
<p>The first major contribution of this thesis is a set of formulations for safety regulators that do not depend on complete knowledge of the underlying dynamical systems. Three unique formulations are proposed, whose usages depend on the level of knowledge of the underlying system. The resulting performance and safety guarantees are analyzed in real-world applications of quadrotor collision avoidance and fast-food frying with industrial manipulators. The second major contribution is a set of two safety filtering frameworks that utilize knowledge of the full-order dynamics, but allow for guaranteed safety in the presence of input constraints on high-dimensional systems. Two formulations are given, with one designed for use on microcontrollers with minimal computational resources. Both formulations utilize the knowledge of an existing "backup controller" that attempts to take the system into a small, safe "backup set". This method is demonstrated in simulation on a robotic manipulator and a Segway robot, and on hardware for collision avoidance and geofencing of single and multi-agent racing drones. The third major contribution is a novel discrete-time formulation of control barrier functions that allow for safety regulation of discrete-time systems. We show how safety constraints can be encoded as temporal logic specifications that are enforced over discrete-time models of the systems and their environments. The fourth and final major contribution is a unified, multi-rate control framework that guarantees safety at both the high-level, in discrete-time, and the low-level, in continuous-time. A mid-level Model Predictive Controller (MPC) is used to generate reference signals based on the high-level planner which are tracked by the low-level controller.</p>
<p>Together, these four major contributions result in safe input regulation on a wide variety of robotic systems. Since no single method can reliably enforce safety on such a wide range of systems with different requirements, this thesis provides the smallest collection of methods that applies to the largest classes of systems.</p>https://resolver.caltech.edu/CaltechTHESIS:06022022-064735213Axial Descent of Multirotor Configurations -- Experimental Studies for Terrestrial and Extraterrestrial Applications
https://resolver.caltech.edu/CaltechTHESIS:01252022-055518852
Year: 2022
DOI: 10.7907/w49w-qy54
<p>Axial descent, specifically the vortex ring state (VRS), poses great challenges for rotorcraft operation as this flight stage is typically accompanied by severe aerodynamic losses and excessive vibrational loads due to the re-ingestion of rotor downwash. Given the hazardous nature of this flight stage, its fluid dynamic properties in regards to single, large-scale rotors have been extensively investigated since the early stages of manned helicopter flight. In light of the rapidly expanding use of small-scale multirotor systems, the field of VRS research has recently received increased interest, with a shifted focus towards small-scale rotors, as the thrust generation and stability of these aerial systems have also been shown to be adversely affected by complex descent aerodynamics. While experimental studies have started examining low Reynolds number rotor aerodynamics in steep or vertical descent, the influence of small-scale rotor geometry and aerodynamic coupling between neighboring rotors have not yet been sufficiently explored.</p>
<p>The objective of this work is, therefore, to extend the current understanding of rotorcraft vortex ring state aerodynamics to low Reynolds number multirotor systems. A series of experimental studies employing various wind tunnel setups and flow visualization techniques is presented with the aim of identifying the underlying ﬂuid-structure interactions, and quantifying rotor performance losses during multirotor axial descent. The work is divided into two fundamental experimental approaches, one utilizing statically mounted rotor systems and one utilizing free-flight testing.</p>
<p>The first part of this work (Chapters 4 and 5) presents the results of wind-tunnel tested statically-mounted rotors for precise aerodynamic identification of rotor performance under simulated descent conditions. Chapter 4 covers a parametric analysis to comprehensively assess the extent to which relevant geometric parameters of a small-scale rotor influence its descent characteristic. Chapter 5 then explores the influence of separation between rotors and identifies potential rotor-rotor interactions in the VRS. The studies in this part of the thesis also make use of PIV setups for visualizing the flow field around small-scale rotors in the axial descent regime, subject to changing geometric parameters and rotor separation.</p>
<p>In the second part (Chapters 6 and 7), a series of free-flight investigations is described for realistically simulated axial descent scenarios. Chapter 6 introduces the methodology for quantifying thrust generation of a multirotor in free-flight without rigid attachment to a load cell, and presents the results of exploratory axial flight studies. Chapter 7 discusses a study on axial descent of variable-pitch multirotor configurations, which was carried out to evaluate the feasibility of deploying a future Mars helicopter in mid air. Findings from this study helped to inform the entry descent and landing (EDL) strategy for JPL's future Martian rotorcraft missions.</p>https://resolver.caltech.edu/CaltechTHESIS:01252022-055518852Nanophotonic Application to Biomedical Devices
https://resolver.caltech.edu/CaltechTHESIS:02182022-230421298
Year: 2022
DOI: 10.7907/tzpw-pt75
<p>Nanophotonics is the study of interactions between nanoscale structures and light. It has greatly expanded the fields of application over the past decades, taking advantage of the advancement in MEMS technology. The most common nanophotonic structures consist of either dielectrics, metals, or both. When a nanophotonic structure contains metals, it is considered as a plasmonic structure. Plasmonics is a field of light-metal interactions. Due to the negative permittivity of metals, the electromagnetic energy of light is focused at the metal-dielectric interface and creates plasmons-a collective motion of electrons in the conduction band of metals. By shaping metals into different structures to achieve a desired performance, plasmonics have been successfully applied to many fields including photovoltaics, spectroscopy, and biomedical devices.</p>
<p>This thesis provides 3 different applications of biomedical devices in which nanophotonics-articularly plasmonics-was applied. Chapter 1 discusses the application of nanophotonics to molecular sensing. In this chapter, an open-top, tapered waveguide that serves as a 3-dimensional plasmon cavity is demonstrated and achieves a near or single molecular detection. Chapter 2 discusses the application of nanophotonics to an implantable intraocular pressure sensor. In this chapter, an array of gold nanodots are introduced on a flexible membrane to optimize the performance of the sensor. Chapter 3 discusses the application of nanophotonics to angle-and-polarization independent pressure or strain sensing, which reduces the need for precise alignment or a trained technician, and therefore can be easily applied to moving subjects in diverse environments. Inspired by the geometry and optical principles of butterfly corneas, an array of gold paraboloids is designed to support a surface plasmon resonance that is angle-and-polarization independent. This array is integrated onto a hermetically sealed cavity with a flexible membrane and enables angle-and-polarization independent pressure/strain sensing.</p>https://resolver.caltech.edu/CaltechTHESIS:02182022-230421298Reliable Controller Synthesis: Guarantees for Safety-Critical System Testing and Verification
https://resolver.caltech.edu/CaltechTHESIS:06122023-162907795
Year: 2023
DOI: 10.7907/jej3-4444
<p>The well-known quote by George Box states that "All models are wrong, but some are useful", and the controls and robotics communities alike have followed a similar paradigm to make significant theoretical and practical advances in the study of controllable systems to date. However, recent robotic system requirements include formal considerations for system safety, especially as we engineer systems that are required to work alongside us in our daily lives. As such, current research directions require analyses that consider these inaccurate system models, our inaccurate understanding of the environments in which these systems operate, and their combined effects on safe, effective system operation, e.g. the canonical autonomous driving problem in exceedingly difficult-to-model urban environments. Recently, this has led to burgeoning efforts in a formal study of controller verification. Specifically, verification denotes the process of determining whether a controller steers its system to exhibit desired behaviors despite the variety of environments the system might face during operation, e.g. whether the autonomous car's controller successfully drives the car to a destination without crashing into obstacles or pedestrians along the way. However, formalization of such a verification pipeline has proved difficult to date, especially since both the models we use for controller synthesis and our understanding of system environments are typically inaccurate.</p>
<p>As a result, this thesis describes our efforts in the development of a formal verification pipeline that addresses a few key challenges in traditional approaches to safety-critical system verification. The first contribution centers on difficult, reactive test synthesis. By test synthesis, we mean the construction of a (potentially difficult) environment in which we require the system under test to perform its objective, e.g. placement of parked cars around which an autonomous vehicle must park. Typically phrased as an optimization problem over the space of allowable environments, these tests are "static" insofar as they do not react to the system's choices made during the test. We posit that such reactivity could more accurately identify worst-case system behavior. As a result, we phrase reactive, maximally difficult test synthesis as a game-theoretic optimization problem, leveraging the same control theoretic tools that facilitate safety-critical controller synthesis - control barrier functions and signal temporal logic. We prove that our proposed synthesis technique is always solvable and always produces a realizable test environment. Finally, we showcase our results by synthesizing reactive tests for both single and multi-agent systems.</p>
<p>The second set of contributions centers on our efforts in uncertainty quantification. Due to un-modeled system and environmental aspects affecting system evolution in unpredictable ways, real-life systems need not realize the same paths every time. As such, typical analyses phrase verification as an optimization problem minimizing the expected value of a function over system trajectories with the expectation taken over this path variability, the distribution for which is assumed to be known. However, we posit that such an analysis should be risk-aware, i.e. account for this variability in a more principled fashion than an expectation-specific analysis, and should not assume apriori knowledge of the distribution corresponding to path variability, as it will be unknown in practice. To that end, we develop methods to bound a subset of risk measures for random variables whose distributions are unknown. This subset includes both Value-at-Risk and other, coherent risk measures heavily utilized in the controls and robotics communities. Simultaneously, we note that the same procedure can be applied to a wide class of non-convex optimization problems. In doing so, we develop a percentile-based optimization approach that rapidly identifies percentile solutions to optimization problems, i.e. a 90-th percentile solution is as good as 90% of solutions in the considered decision space.</p>
<p>The third set of contributions focuses on the application of the prior mathematical developments to facilitate both risk-aware safety-critical system verification and controller synthesis. We phrase risk-aware controller verification as a risk-measure identification problem and utilize the prior bounding results to provide an efficient, dimensionally-independent verification procedure. Then, we phrase risk-aware controller synthesis as an optimization problem maximizing the bound provided by our risk-aware verification method, and show that this problem is solvable by the percentile optimization methods mentioned prior. Finally, we lay the foundation for the utilization of the aforementioned mathematical developments in other aspects of controls and robotics and communities more broadly. We show how risk-measure bounding can augment models both offline and online to robustify safety-critical controllers, how percentile optimization can facilitate "optimal" input selection and guarantee generation for non-convex finite-time optimal controllers, and how multiple applications of the percentile approach can also bound the optimality gap of reported percentile solutions. We showcase all these results on hardware for multiple systems and highlight the data efficiency of our proposed approaches.</p>https://resolver.caltech.edu/CaltechTHESIS:06122023-162907795Robust Safety-Critical Control: A Lyapunov and Barrier Approach
https://resolver.caltech.edu/CaltechTHESIS:06022023-032907616
Year: 2023
DOI: 10.7907/bpht-by81
<p>Accompanying the technological advances of the past decade has been the promise for widespread growth of autonomous systems into nearly all domains of human society, including manufacturing, transportation, and healthcare. At the same time, there have been several tragic failures that reveal potential risks with the expansion of autonomous systems into everyday life, and indicate that it is vital for safety to be accounted for in the design of control systems.</p>
<p>This thesis seeks to develop a theory of robust safety-critical control for autonomous systems. This theory will be built upon the foundational tools of Control Lyapunov Functions (CLFs) and Control Barrier Functions (CBFs), which provide a powerful paradigm for the design of model-based safety-critical controllers. The dependence of CLF and CBF-based controllers on a system model makes them susceptible to modeling inaccuracies, potentially resulting in unsafe behavior when deploying these controllers on real-world systems.</p>
<p>In this thesis I present methods for resolving four classes of model inaccuracies referred to as model error, disturbances, measurement error, and input sampling, which are commonly faced challenges when designing controllers for robotic systems. The proposed methods are unified by their shared use of CLFs and CBFs to produce controllers possessing rigorous and robust safety guarantees that can be demonstrated in simulation or experimentally. A hallmark of these methods is a focus on enabling control synthesis through convex optimization, which ensures that controllers can be efficiently computed on real-world robotic hardware platforms.</p>
<p>In addressing model error, I consider both data-driven learning approaches and adaptive control approaches. I present three episodic learning frameworks that iteratively augment existing CLF and CBF-based controllers specified via convex optimization problems to improve the stability and safety properties of a system, which I demonstrate in simulation and experimentally. I also establish a relationship between the degradation of stability and safety properties with the magnitude of residual learning error through the perspective of Input-to-State Stability (ISS) and Input-to-State Safety (ISSf). Lastly, I develop an adaptive safety-critical control framework for systems with parametric model error through the notion of adaptive CBFs.</p>
<p>In addressing disturbances, I resolve challenges in balancing performance and robustness with ISSf-based controllers through the notion of Tunable Input-to-State Safety (TISSf), which permits prioritizing robustness to disturbances only when safety requirements are close to being violated. I demonstrate the capabilities of TISSf-based control design experimentally on an autonomous semi-trailer truck system that is subject to input disturbances due to complex unmodeled actuator dynamics. Lastly, I develop a framework for achieving ISSf-like finite-time safety guarantees for discrete-time systems subject to stochastic disturbances through the use of CBFs and convex optimization.</p>
<p>In addressing measurement error, I develop the notion of Measurement-Robust CBFs (MR-CBFs), which permit control synthesis through convex optimization in the presence of imperfect measurements. I demonstrate the capability of MR-CBFs on an experimental Segway system using a vision-based measurement system, validating the tractability of using controllers specified through increasingly complex classes of convex optimization problems on real-world systems. Lastly, I present an application of Preference Based Learning (PBL) in tuning the robustness parameters of a CBF-based controller, demonstrating the first use of PBL with CBFs and providing a tool for tuning the safety and performance of the robust controllers proposed in this thesis.</p>
<p>In addressing input sampling, I consider both sampled-data and event-triggered paradigms for modeling input sampling. I provide a method for synthesizing CLF-based controllers for sampled-data systems by integrating feedback linearization with approximate discrete-time models, leading to a significant improvement over continuous-time CLF-based controllers implemented with input sampling. I then develop a framework for achieving safety of sampled-data systems through approximate discrete-time models through the notion of practical safety and Sampled-Data CBFs (SD-CBFs), which I demonstrate with convex-optimization based controllers in simulation. Lastly, I develop a method for event-triggered safety-critical control that uses ISSf to achieve safety while satisfying the requirement of a minimum interevent time.</p>
<p>Collectively, these contributions constitute a significant advance in the theory of robust safety-critical control by establishing a framework, unified by the use of CLFs and CBFs in conjunction with convex optimization, that addresses a wide class of challenges faced in the design of safety-critical control systems.</p>https://resolver.caltech.edu/CaltechTHESIS:06022023-032907616A Neural Network Model of an Insect's Wing Hinge Reveals How Steering Muscles Control Flight
https://resolver.caltech.edu/CaltechTHESIS:02272023-213525351
Year: 2023
DOI: 10.7907/teej-tb66
<p>The flight system of the fly is remarkable. A fly can execute an escape maneuver in milliseconds, compensate for wing damage when half of the wing is missing, fly in turbulent conditions, and migrate over large distances. While there are many factors that contribute to the robustness and versatility of insect flight, it is the mechanical encoding of wing motion in the wing hinge that allows flies to rapidly and accurately change wing motion over a large dynamic range. The wing hinge consists of several hardened skeletal elements, named sclerites, and a set of twelve steering muscles are attached to some of these components within the exoskeleton. Due to the anatomical complexity and minute size of the sclerites, the way in which the steering muscles alter the mechanical encoding of wing motion in the hinge is poorly understood.</p>
<p>Using genetically encoded calcium indicators and high-speed videography, is is possible to simultaneously image steering muscle activity and wing motion. In order to extract wing pose from the high-speed video frames, an automated tracking algorithm was developed, that used a neural network and model fitting to accurately reconstruct the wing kinematics. The synchronous recordings of wing motion and steering muscle activity were used to train a convolutional neural network that learned to accurately predict the wing kinematics from muscle activity patterns. After training, the convolutional neural network was used to perform virtual experiments, revealing how the steering muscles regulate wing motion. Correlation analysis revealed that the 12 steering muscles have highly correlated activity. The correlation of muscle activity can be approximated well by a 12D-plane, in which all activity has to reside.</p>
<p>To study the function of the sclerites, a bottleneck was introduced in the convolutional neural network. The bottleneck consists of five neurons, or latent parameters, four parameters corresponding to the state of the different sclerites, on which the steering muscles act, and one parameter representing the wingbeat frequency. This so called latent network predicts both the changes in wing motion and muscle activity patterns as a function of sclerite state. The predicted wing motion as a function of sclerite state matches with previous anatomy and electrophysiology studies for the basalare, first axillary and third axillary sclerites. The fourth axillary sclerite has not been studied before, but shows an antagonistic relationship between the hg<sub>1,2</sub> and hg<sub>3,4</sub> muscles, resulting in a strong decrease and increase, respectively, of stroke amplitude, deviation and wing pitch angles.</p>
<p>By replaying the wing kinematics of the virtual experiments on a dynamically scaled robotic fly, a model of the aerodynamic and inertial control forces as a function of steering muscle activity was constructed. This control force model was subsequently integrated in a state-space system of fly flight, which in turn was integrated in a model predictive control simulation that was used to simulate free flight maneuvers. The body motion, steering muscle activity, and wing kinematics of the model predictive control simulations were strikingly similar to the recorded maneuvers of free-flying flies.</p>
<p>The integrative, multi-disciplinary approach that was used to reveal the mechanical logic of the wing hinge, and the control problem that a fly needs to solve to stay airborne, are both unprecedented in prior literature. The methodologies and models of this study will be a valuable resource in future research on how the fly's nervous system controls the complex behavior that is flight.</p>https://resolver.caltech.edu/CaltechTHESIS:02272023-213525351Risk-Aware Planning and Control in Extreme Environments
https://resolver.caltech.edu/CaltechTHESIS:02082023-223824752
Year: 2023
DOI: 10.7907/xv2b-tj24
<p>Safety-critical control and planning for autonomous systems operating in unstructured environments is a challenging problem must be addressed as autonomous vehicles, surgical robots, and autonomous industrial robots become more pervasive. This thesis addresses some of the issues in safety critical autonomy by introducing new techniques for computationally tractable and efficient safety-critical control. The approach developed in this thesis arises from taking a deeper look at two questions: 1) How can we obtain better uncertainty quantification of the disturbances that affect autonomous systems either as a result of unmodeled changes in the environment or due to sensor imperfections? 2) Given richer uncertainty quantification techniques, how do incorporate the diverse uncertainty descriptions into the control and planning framework without sacrificing the tractability and efficiency of existing approaches?</p>
<p>I address the above two questions by developing risk-aware control and planning techniques for traversal of a mobile robot over static but extreme terrain and in the presence of dynamic obstacles. We first look at algorithms for risk-aware terrain assessment, and extensively test them on wheeled and legged robots that were deployed in subterranean tunnel, urban, and cave environments for search and rescue operations in the DARPA Subterranean Challenge. I then present a theory for risk-aware model predictive control in static environments and in the presence of dynamic obstacles. Coherent risk measures are applied to this planning and control framework in order to account for diverse uncertainty descriptions. Computationally tractable reformulations of the optimal control problem are realized through constraint tightening techniques.</p>
<p>I then investigate algorithms for uncertainty assessment and prediction of apriori unknown, dynamic obstacles using data-driven techniques. We use a technique from signal processing literature called Singular Spectrum Analysis for making linear predictions of dynamic obstacles. The obstacle motion predictions are equipped with error predictions to account for the uncertainty in the sensing heuristically using bootstrapping techniques. We use a statistical tool, Adaptive Conformal Inference, to further calibrate the heuristic error prediction online to obtain true uncertainty prediction while using nonstationary data to analyze the performance of the data-driven predictor. These techniques provide reactive, real-time, risk-aware obstacle avoidance in dynamic environments.</p>https://resolver.caltech.edu/CaltechTHESIS:02082023-223824752Reliable Learning and Control in Dynamic Environments: Towards Unified Theory and Learned Robotic Agility
https://resolver.caltech.edu/CaltechTHESIS:08052022-231458463
Year: 2023
DOI: 10.7907/8rz4-7b35
<p>Recent breathtaking advances in machine learning beckon to their applications in a wide range of real-world autonomous systems. However, for safety-critical settings such as agile robotic control in hazardous environments, we must confront several key challenges before widespread deployment. Most importantly, the learning system must interact with the rest of the autonomous system (e.g., highly nonlinear and non-stationary dynamics) in a way that safeguards against catastrophic failures with formal guarantees. In addition, from both computational and statistical standpoints, the learning system must incorporate prior knowledge for efficiency and generalizability.</p>
<p>This thesis presents progress towards establishing a unified framework that fundamentally connects learning and control. First, Part I motivates the benefit and necessity of such a unified framework by the Neural-Control Family, a family of nonlinear deep-learning-based control methods with not only stability and robustness guarantees but also new capabilities in agile robotic control. Then Part II discusses three unifying interfaces between learning and control: (1) online meta-adaptive control, (2) competitive online optimization and control, and (3) online learning perspectives on model predictive control. All interfaces yield settings that jointly admit both learning-theoretic and control-theoretic guarantees.</p>https://resolver.caltech.edu/CaltechTHESIS:08052022-231458463Enabling Robust and User-Customized Bipedal Locomotion on Lower-Body Assistive Devices via Hybrid System Theory and Preference-Based Learning
https://resolver.caltech.edu/CaltechTHESIS:04292023-003436131
Year: 2023
DOI: 10.7907/j9hk-xa17
<p>Practical robotic assistive devices have the potential to transform many aspects of our society, from enabling locomotive autonomy to facilitating rehabilitation. However, as is typically the case when having autonomous systems interact closely with humans, one must simultaneously solve multiple grand challenges. My work focuses specifically on 1) leveraging hybrid system theory to achieve stable and robust walking that generalizes well across various human models and environmental conditions, and 2) developing an online learning strategy to customize the experimental walking for individual user comfort. The presented methodology is grounded in realizing lower-body exoskeleton locomotion for subjects with motor complete paraplegia, with extensions to other robotic applications. The contributions are broken down as follows.</p>
<p>First, by leveraging tools from nonlinear control theory, I propose techniques for systematically addressing locomotive robustness. These techniques include: using saltation matrices to generate robust gaits with experimental demonstrations on the Atalante lower-body exoskeleton; and developing an input-to-state stability perspective to certify robustness to uncertain impact events. Importantly, these methods aim to better understand the mathematical conditions underlying robust locomotion -- a necessary step towards realizing safe locomotion across varying human models and environmental conditions. Second, I develop a preference-based learning framework to explicitly optimize user comfort during exoskeleton locomotion (achieved using the aforementioned nonlinear control methodology) by learning directly from subjective feedback. This framework is implemented in real-world settings, including the clinical realization of user-preferred locomotion for two subjects with motor complete paraplegia.Third, the extensibility of this framework is demonstrated through three general robotic applications: tuning constraints of the gait generation optimization problem with demonstrations on a planar biped; tuning Lyapunov-based controller gains on a 3D biped; and tuning control barrier function parameters for performant yet safe exploration on a quadrupedal platform. Lastly, I discuss other relevant clinical considerations for lower-body assistive devices including how exoskeleton locomotion influences metabolic cost of transport, the study of latent factors underlying user-preferred walking, and embedding musculoskeletal models directly in the gait generation process.</p>https://resolver.caltech.edu/CaltechTHESIS:04292023-003436131Studies on Off-Nominal Rotor Aerodynamics for eVTOL Aircraft
https://resolver.caltech.edu/CaltechTHESIS:12222022-065507477
Year: 2023
DOI: 10.7907/eytr-nd50
<p>As electric Vertical Takeoff and Landing (eVTOL) aircraft become increasingly common, improved understanding of rotor aerodynamics in off-nominal conditions becomes ever more important. A better fundamental understanding of these effects can help inform vehicle design, leading to lower power consumption and improved performance. This thesis will cover a selection of topics to gain a better understanding of the expected rotor aerodynamics associated with use in this class of vehicle, as well as the development of tools to aid in the studies and an analysis of the impact of the effects.</p>
<p>To consider special effects on a rotor in hover on such a vehicle, Chapter 2 is the study of obstructions in the upstream of a propeller, representing the effects of a wing or fuselage blocking a propeller’s inlet. The next is the effect of forward flight on the forces produced by a rotor. Lifting rotors are often used in eVTOL aircraft as the craft transitions to forward flight, so a study of their performance in forward flight as well as a model are presented in Chapter 3. Having examined rotor-wing interactions in hover and isolated rotor performance in forward flight, the next step is to examine rotor-wing interactions in forward flight. Chapter 6 shows the design of an integrated test stand for studying the aerodynamic interactions between lifting propellers and a wing in low-speed, transitional forward flight, as well as the subsequent results.</p>
<p>This thesis also describes the development and implementation of two tools to aid in the work herein. The first (Chapter 4) is a rapid, low-cost method of extracting the geometry of a propeller using photogrammetry which is subsequently used in simulations. The second (Chapter 5) is low-cost and accessible multi-axis force sensor used in the integrated test stand for propeller-wing interaction studies. To assess the impact of the findings, the experimental results and models developed are then taken into consideration by applying them to models of existing eVTOL aircraft in Chapter 7. The change in modeling of hover and transition performance is studied with and without the additional modeling.</p>https://resolver.caltech.edu/CaltechTHESIS:12222022-065507477Model-Based Lower-Limb Powered Prosthesis Control: Developing and Realizing Nonlinear Subsystem Control Methods for Generalizable Prosthesis Control
https://resolver.caltech.edu/CaltechTHESIS:01072023-214003146
Year: 2023
DOI: 10.7907/6724-6e14
<p>While there are over 600,000 lower-limb amputees in the US, commercially available prostheses remain limited to mostly passive devices. People that walk with a passive prosthesis experience an increase in energy expenditure, a decrease in comfortable walking speed, and gait asymmetry which leads to degenerative conditions. To address these limitations, researchers have developed powered prostheses with the aim of replicating the net positive energy biological limbs supply to humans in walking. These active devices have been shown to decrease users' metabolic cost and increase their comfortable walking speed. However, the control methods to achieve these results typically require hours of heuristic tuning for every user and every behavior. This motivates developing more formal prosthesis control methods that generalize between users.</p>
<p>Formal nonlinear control methods have been developed to realize energy efficient, human-like walking on bipedal robots. These model-based approaches provide a systematic approach to generate and realize provably stable walking gaits. However, these methods cannot be directly applied to prostheses since they depend on a dynamic model of the entire system, and in the case of the prosthesis, the human dynamics are unknown.</p>
<p>To address this challenge, we develop a theoretical framework to translate model-based bipedal control methods to prostheses with the aim of realizing a generalizable prosthesis control method. We separate the prosthesis subsystem from the remaining human portion of the system and model the human's impact on the prosthesis dynamics with a measure of the interaction forces between the human and the prosthesis. We theoretically prove that a model-based controller developed in this separable subsystem framework is equivalent to one developed with knowledge of the full-order human-prosthesis system. With control Lyapunov functions, we develop a wider class of subsystem controllers that solely depend on local information but provide full-order system guarantees, even in the presence of force estimate errors. This work bridges the gap between bipedal control methods and prostheses, allowing us to leverage the benefits of model-based approaches on prostheses.</p>
<p>We demonstrated a controller of this class through an online optimization-based approach on a powered knee-ankle prosthesis, realizing the first model-dependent lower-limb prosthesis controller that accounts for the interaction force between the human and the prosthesis. For a first pass, a force-estimation method was used that yields improved tracking of the desired trajectories over model-independent prosthesis control methods. Then, we incorporated a load cell into the prosthesis platform at the human-prosthesis attachment point to measure the interaction forces, and an inertial measurement to measure the rotation and velocity of the human's thigh. These two sensors completed the prosthesis dynamics model. A pressure sensor incorporated into the prosthesis' shoe measured the ground reaction forces, enabling the prosthesis to respond to its real-world environment, proving robust to 4 different terrains. We extended this controller to a multi-domain hybrid system approach to model the changing contact points occurring in human heel-toe roll. By allowing the prosthesis to sense the human's large varying dynamic load and respond accordingly, this model-based prosthesis controller emulated subject-specific human kinematic trends on a knee-ankle prosthesis for two subjects with no tuning in between, suggesting this approach could yield a method that generalizes between users.</p>https://resolver.caltech.edu/CaltechTHESIS:01072023-214003146Methods for Robust Learning-Based Control
https://resolver.caltech.edu/CaltechTHESIS:06072023-134620248
Year: 2023
DOI: 10.7907/2xnc-t162
<p>This thesis addresses the general problem of improving control, safety, and reliability of multi-rotor drones in various challenging conditions by introducing novel deep-learning-based approaches. These approaches are designed to tackle specific issues that multi-rotor drones face during operation, such as near-ground trajectory control, high-speed wind disturbances, actuation delays, and motor failures. The thesis is organized into four main chapters, plus an introduction and conclusion. Each of the main chapters focuses on a unique approach to address a particular challenge of deep-learning-based control methods. Chapter 2 presents Neural-Lander, a deep-learning-based robust nonlinear controller that significantly improves quadrotor control performance during landing by accounting for complex aerodynamic effects. This chapter addresses key challenges to incorporating learned residual dynamics into a control architecture, laying the groundwork for the subsequent chapters. Chapters 3 and 4 introduce Neural-Fly, a learning-based approach that uses Domain Adversarially Invariant Meta-Learning (DAIML) and adaptive control to enable rapid online learning and precise flight control under a wide range of wind conditions. Chapter 5 proposes a lightweight augmentation method that enhances trajectory tracking performance for UAVs by effectively compensating for motor dynamics and digital transport delays. This method is extensible to a range of control methods, including learning-based approaches. Chapter 6 explores a novel sparse failure identification method for detecting and compensating for motor failures in over-actuated UAVs, contributing to the development of robust fault detection and compensation strategies for a safer and more reliable operation. This method builds on the Neural-Fly online learning framework and extends it to handle a wider range of conditions, including complete actuator failures. Together, these chapters address key challenges in safe and reliable learning-based control and demonstrate the potential of deep-learning-based control methods.</p>https://resolver.caltech.edu/CaltechTHESIS:06072023-134620248Control of Unknown Dynamical Systems: Robustness and Online Learning of Feedback Control
https://resolver.caltech.edu/CaltechTHESIS:09062023-095903699
Year: 2024
DOI: 10.7907/fb64-vk24
<p>Over the past few decades, our physical and digital worlds have become increasingly intertwined and reliant on each other. Advancements in areas such as machine learning, online optimization, and control theory, along with ubiquitous access to computational power, have played a crucial role in this technological evolution. As a result, we are now moving towards a future where complex and intelligent dynamical systems, with humans in the loop, govern our daily lives.</p>
<p>Building advanced control systems is a critical step in this journey, as they enable swift and data-informed decision-making. However, as we aim to create even more sophisticated closed-loop systems, we must proceed with a careful balance of ambition and caution. While the benefits of these interconnected systems are abundant and our dependence on them deepens, ensuring the actual reliability and safety of the systems becomes increasingly challenging due to the growing complexity of their dynamics. This challenge is particularly prominent in safety-critical applications involving physical systems, which often have strict and non-negotiable safety and performance requirements. To establish a harmonious relationship between our physical and digital worlds, it is crucial to develop intelligent closed-loop control systems that are not only fast and efficient, but also reliable and fault-tolerant.</p>
<p>The title of this thesis, "Control of Unknown Dynamical Systems: Robustness and Online Learning of Feedback Control," reflects the central focus of this work on addressing this pressing challenge. The thesis aims to develop theoretical frameworks and tools that provide insights and contribute new approaches to the design of control systems capable of handling the inherent uncertainty in real-world dynamical systems.</p>
<p>The first part of the thesis focuses on the design of closed-loop systems that are robust to dynamic uncertainty, particularly in settings involving nonlinear dynamics and complex control constraints. The second part introduces a general framework for learning-to-control algorithms that provide worst-case guarantees, even in scenarios where the dynamic uncertainty is arbitrarily large. By addressing these key aspects, this work aims to advance our understanding and capabilities in designing control systems that can effectively deal with uncertainty.</p>https://resolver.caltech.edu/CaltechTHESIS:09062023-095903699Formal Methods for Test and Evaluation: Reasoning over Tests, Automated Test Synthesis, and System Diagnostics
https://resolver.caltech.edu/CaltechTHESIS:05312024-094443866
Year: 2024
DOI: 10.7907/4xdc-b988
<p>With the integration of autonomous systems into our everyday lives edging closer to reality, ensuring the safety of these systems is paramount. Part of the safety verification process is a rigorous testing procedure, which currently does not exist for autonomous vehicles. In this thesis, we aim to provide approaches using formal methods to increase the efficiency of testing campaigns.
First, we provide a framework based on assume-guarantee contracts to specify tests in the form of a test structure. Using these test structures, we then show how to combine, split, and compare tests. Additionally, we characterize when tests can be combined and when the resulting test requires temporal constraints. Next, we demonstrate the approach on examples and find a strategy for a test agent using winning sets and Monte Carlo tree search.</p>
<p>Second, we present a framework to automatically synthesize a test environment, consisting of static and reactive obstacles, and dynamic test agents. We characterize the desired test behavior in a system and a test objective in the form of a linear temporal logic specification, consisting of sub-tasks commonly used for robotic missions. This test environment must ensure that the test is not impossible (i.e. a correct system can pass the test), but also that every test execution that satisfies the system objective also satisfies the test objective. We use tools from automata theory to construct the virtual product graph that represents all possible test executions, and the virtual system graph, which corresponds to the system's perspective.
We formulate this routing problem as a network flow optimization on the virtual product graph in the form of a mixed integer linear program for different test environments. We show that this routing problem is NP-hard. We propose a counterexample-guided search using GR(1) synthesis to find a strategy for a test agent. This framework is demonstrated in several examples in simulation and hardware.</p>
<p>Lastly, we present a framework to diagnose a system-level fault by identifying the component responsible for the failure. We make use of assume-guarantee contracts and Pacti, a tool for compositional system analysis and design, to construct a diagnostics map, which allows us to trace a system-level guarantee to possible causes. We show that this framework can reduce the number of statements that need to be checked in the diagnostics process. We illustrate this framework on several abstract examples and two examples inspired by a real-world autonomous system.</p>https://resolver.caltech.edu/CaltechTHESIS:05312024-094443866Test and Evaluation of Autonomous Systems: Reactive Test Synthesis and Task-Relevant Evaluation of Perception
https://resolver.caltech.edu/CaltechTHESIS:06022024-014038700
Year: 2024
DOI: 10.7907/e8qz-rd26
<p>Autonomous robotic systems have potential for profound impact on our society -- legged and wheeled robots for search and rescue missions, drones for wildfire management, self-driving cars for improving mobility, and robotic space missions for exploration and repair of spacecraft. The complexity of these systems implies that formal guarantees during the design phase alone is not sufficient; mainstream deployment of these systems requires principled frameworks for test and evaluation, and verification and validation. This thesis studies two such challenges to mainstream deployment of these systems.</p>
<p>First, we consider the problem of evaluating perception models in a manner relevant to the system-level specification and the downstream planner. Perception and planning modules are often designed under different computational and mathematical paradigms. This talk will focus on evaluating models for classification and detection tasks, and leverages confusion matrices which are popularly used in computer vision to evaluate object detection models to derive probabilistic guarantees at the system-level. However, not all perception errors are equally safety-critical, and traditional confusion matrices account for all objects equally. Thus, task-relevant metrics such as proposition labeled confusion matrices are introduced. These are constructed by identifying propositional formulas relevant to the downstream planning logic and the system-level specification, and result in less conservative system-level guarantees. Using this analysis, fundamental tradeoffs in perception models are reflected in the tradeoffs of probabilistic guarantees. This framework is illustrated on a car-pedestrian example in simulation, and the confusion matrices are constructed from state-of-the-art detection models evaluated on the nuScenes dataset.</p>
<p>Second, we consider the problem of automatically synthesizing tests for autonomous robotic systems. These systems reason over both discrete (e.g., navigate left or right around an obstacle) and continuous variables (e.g., continuous trajectories). This talk presents a flow-based approach for test environment synthesis which handles discrete variables and is also reactive to the system under test. Reactivity is important to account for uncertainties in system modeling, and to adapt to system behavior without knowledge of the system controller. These tests are synthesized from high-level specifications of desired behavior. Though the problem is shown to be NP-hard, a flow-based mixed-integer linear program formulation is used that scales well to medium-sized examples (e.g., >10,000 integer variables). The test environment can consist of static and reactive obstacles as well as dynamic test agents, whose strategies are synthesized to match the solution of the flow-based optimization. The overview of the approach is as follows. First, principles of automata theory are used to translate the high-level system and test objectives, and the non-deterministic abstraction of the system into a network flow optimization. The solution of this optimization is then parsed into GR(1) formulas in linear temporal logic. This GR(1) formula is used to synthesize reactive strategies of a dynamic test agent in a counterexample-guided fashion. We provide guarantees that the synthesized test strategy will realize the desired test behavior under the assumption of a well-designed system, the test strategy is reactive and least-restrictive,. This framework is illustrated on several simulation and hardware experiments with quadrupeds, showing promise towards a layered approach to test and evaluation.</p>https://resolver.caltech.edu/CaltechTHESIS:06022024-014038700Autonomous Flow-Based Navigation in Unsteady Underwater Environments
https://resolver.caltech.edu/CaltechTHESIS:06052024-052757779
Year: 2024
DOI: 10.7907/vnh6-3t44
Autonomous ocean-exploring robots promise to significantly enhance the rate at which we can explore ocean environments. However, the limited range and speed of existing autonomous underwater vehicles (AUVs) are barriers to comprehensive ocean exploration. To address these limitations, the work in this thesis investigates strategies for improving the capabilities of existing AUVs, such as targeted sampling and efficient navigation through background flows. Inspired by the ability of aquatic animals to navigate via flow sensing, hydrodynamic cues are investigated as a sensory input for accomplishing these feats of autonomous navigation using only onboard sensors. First, reinforcement learning (RL) is investigated as an algorithm for accomplishing efficient point-to-point navigation in simulated cylinder flow. The algorithm entails inputting point measurements of flow quantities such as velocity and vorticity into a deep neural network, which then determines a swimmer's actions. Using point velocity as the sensory input, the RL algorithm achieved a near 100 percent success rate in reaching the target locations while approaching the time-efficiency of optimal navigation trajectories. To test RL and flow-based navigation in a physical setting, we next developed the Caltech autonomous reinforcement learning robot (CARL), a palm-sized underwater robotic platform. As proof-of-concept analogy for tracking hydrothermal vent plumes in the ocean, the robot was tasked with locating the center of turbulent jet flows in a 13,000-liter water tank using data from onboard pressure sensors. Using a navigation policy trained with RL in a simulated flow environment, CARL successfully located the turbulent plumes at more than twice the rate of random searching by detecting mean flow gradients with the onboard pressure sensors. Lastly, combing both flow sensing and efficient navigation, the accelerometer onboard CARL was used to sense and exploit the flow from a passing vortex ring for energy-efficient propulsion. Body acceleration and rotation were shown to be effective methods of indirect flow sensing, which enabled the energy-efficient vortex ring surfing strategy. Throughout this work, efforts are made to understand the governing physics behind the discovered navigation strategies to generalize the results beyond a specific navigation problem, sensor type, or robotic implementation.https://resolver.caltech.edu/CaltechTHESIS:06052024-052757779Data-Driven Safety-Critical Autonomy in Unknown, Unstructured, and Dynamic Environments
https://resolver.caltech.edu/CaltechTHESIS:03042024-201031352
Year: 2024
DOI: 10.7907/qpbp-0x81
<p>This thesis addresses the critical challenge of ensuring safety in autonomous exploration within unknown, unstructured, dynamic environments, a domain filled with various types of uncertainties. These include model uncertainties in system dynamics, localization uncertainties stemming from measurement noises, and the risks of collision in environments with dynamic obstacles. Traditional models for vehicle planning and control are often simplified for computational feasibility, but this simplification without careful analysis can compromise safety and system stability. My research introduces a novel, comprehensive framework to provide probabilistically safe planning and control for robot autonomy, structured around three components:</p>
<p>(1) Probabilistic Uncertainty Quantification for Model Mismatches: </p>
<p>This segment focuses on identifying model discrepancies given closed-loop tracking data in an unstructured environment where a reduced-order robot model is used for planning and control. The disturbance is modeled as a scalar-valued stochastic process of a norm on the difference between the reduce-order robot model and actual system evolution. In an online and risk-aware framework, Gaussian Process Regression is employed to extract the probabilistic upper bound to such stochastic process, referred to as the Surface-at-Risk. Theoretical guarantees on the accuracy of the fitted discrepancy surface are analyzed and verified to the data sets collected during system operation. </p>
<p>In an offline setting, conformal prediction, a statistical inference tool, is employed to obtain probabilistic upper bounds of matched and unmatched model disturbance in the system from data, without any assumption of the latent probability distribution governing these discrepancies. Building on these bounds, the robot's nominal ancillary controller is augmented for extending robustness and stability guarantees of the closed-loop system in the face of such discrepancies. Additionally, a maximum tracking error tube is constructed along the planned trajectory using the reduced-order model. Such error tubes describe the maximum permissible deviation in actual trajectory tracking under the augmented ancillary controller and the worst-case matched and unmatched model uncertainties, thereby delineating safe operational boundaries for the system. </p>
<p>(2) Data-Driven Unsafe Set Prediction for Dynamic Obstacles: </p>
<p>This thesis topic develops an online, data-driven predictive model for dynamic obstacles, accounting for measurement noise and low-frequency data rates.
First inspired by singular spectrum analysis (SSA), a time-series forecast technique, obstacle models characterized by linear recurrence relationships are extracted from real-time position observables. Using the statistical bootstrap technique, a set of predicted obstacle trajectories are constructed, which in turn are reformulated into deterministic distributionally robust obstacle avoidance constraints, reflecting a user-defined risk tolerance. </p>
<p>Further refining the obstacle predictor for intention-unknown obstacles, a linear, time-varying model is learned from data using time-delay embedding of obstacle position observables. Additive process and measurement noises are anticipated in the learned model, where their intensities are estimated from data. For inferring prediction uncertainties, a companion data-driven Kalman Filter (DDKF) is constructed to forecast obstacle positions and uncertainties. This "heuristic unsafe set" from DDKF is then dynamically calibrated using adaptive conformal prediction, ensuring safety without relying on any distribution assumptions regarding the uncertainties or model accuracy. The calibrated sets, called conformal prediction sets, are then reformulated into convex state constraints.</p>
<p>(3) Safety-Critical Planning:</p>
<p>The thesis proposes two methods for ensuring safety in planning and navigation: Probabilistic-Safe Model Predictive Control (MPC) and Probabilistic-Safe Model Predictive Path Integral (MPPI) given uncertainties arising from operating in unknown, unstructured, and dynamic environments. The MPC approach integrates the quantified obstacle avoidance constraints into a convex program to balance computational tractability while providing probabilistic safety guarantees. In contrast, the MPPI method, a sampling-based strategy, incorporating unsafe sets into a cost map derived from sensory data, optimizes reference tracking trajectory while guaranteeing collision avoidance up to a user-defined risk tolerance.</p>
<p>In unknown and cluttered environments automatically, the proposed framework learns an upper bound on model residuals from data and systematically calculates the safety buffers needed to provide the desired probabilistic safe navigation of robotics systems. Additionally, in the presence of dynamic obstacles, the proposed data-driven predictor systematically extracts an obstacle model and makes obstacle-occupied unsafe set forecasts. These features largely eliminate the "hand tuning" of the underlying planner and controller that is normally required in heuristic-based algorithms. The efficacy of these proposed frameworks is empirically validated through Monte Carlo Simulations, alongside hardware validations on both ground and aerial vehicles, demonstrating their robustness, versatility, and applicability in real-world scenarios.</p>https://resolver.caltech.edu/CaltechTHESIS:03042024-201031352