Seminars

Seminars

Seminars are held on Tuesdays at 3:00 pm in 206 Thomas. Please e-mail comments, questions, and requests to be added to the seminar e-mail list to Maria Koeper.

July 27, 2007 (Thesis Seminar)
Yongqiang Liang
Robotic Training on Motor Rehabilitation after Complete Spinal Cord Injury
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's isolated from supraspinal nervous systems.

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's 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.

Standing is another important function in one's daily life while 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 respond to postural disturbance. Further studies in standing rehabilitation should combine other methods to provide tonic inputs to the spinal cord.

September 19, 2007 (Thesis Seminar)
Tomonori Honda
Grayscale Reliability in Engineering Design
Every product has inherent failure modes and impacts of those failures. Products also typically have required or desired system reliability specifications. The most readily available reliability data is on component behavior (typically including statistical predictions of performance degradation). Therefore, it would be useful to be able to develop predictions of system degradation and failure based on component statistics. Currently, two distinct approaches are used for analyzing robustness and reliability of system performance by propagating the states of components: probabilistic risk assessment (PRA) and robust design techniques. Probabilistic risk assessment was developed by the nuclear industry for analyzing system reliability when component failures consist of low probability but catastrophic events. On the other hand, robust design techniques, such as Taguchi's method, were developed to analyze systems for performance loss caused by manufacturing and operational variations. However, when analyzing aerospace missions, both the degree of system degradation and the time dependence of system degradation is important.

This thesis presents the development of a new approach for computing and analyzing the continuous (or grayscale) degradation to the system performance from fully working to completely failed states given the degradation profile for its components over the time. Using three examples with different complexity, usefulness of the new approach is explored. Using a simple Mass-Spring-Damper system, the grayscale approach will be demonstrated and the difference between new approach and fault tree analysis is shown. The 2-D Mass-Spring-Damper system example is used to demonstrate how grayscale reliability captures the coupling effect as well as how the designer can use grayscale reliability to improve designs using sensitivity analysis. Finally, a Stirling Engine is given as a realistic example and to show how it captures sensitivity of components to overall system performance.

September 19, 2007 (Thesis Seminar)
Fabien Nicaise
Automated Design Synthesis of Structures Using Growth Enhanced Evolution
Conceptual design is the process of generating and evaluating candidate solutions to a problem without a detailed knowledge of all aspects of that solution or even the problem itself. This is fraught risk as the requirements are still uncertain as is the whole concept. Yet, the decisions made in this phase already determine the eventual fate of design.

Evolutionary Algorithms are a class of computational techniques, based on Darwinian evolution, to search design spaces solutions to complicated problems. These methods have met with great success across many disciplines on diverse types of problems. In mechanical design, however most successful applications of Evolutionary Algorithms are applied to fairly specific problems in the later stages of the design life cycle. The result is that they are used for optimization of a given topology rather than synthesizing new designs.

The work presented here aims to remedy this situation and provide an new encoding scheme for discrete structures to enable better evolution. This encoding scheme, called 'Indirect Encoding', represents the candidate solutions as an instruction set on building the individuals rather than directly representing the structure itself. The encoding scheme along with examples of hand coded and evolved solutions will be present to illustrate the method and the benefits.

October 8, 2007 (Thesis Seminar)
Tamer Elsayed
306 Firestone
A Variational Constitutive Model for Polymers and Soft Biological Tissues with Applications
Soft materials exhibit complex mechanical behavior, characterized by large strains, hysteresis, rate-sensitivity, stress softening (Mullins effect), deviatoric and volumetric plasticity. The need to accurately predict the behavior of such materials for the purpose of practical applications has been a tremendous challenge for scientists and engineers. 

This work presents a seamless, fully variational constitutive model capable of capturing all of the above complex characteristics. A fitting procedure based on the use of Genetic Algorithms is utilized to identify the necessary material parameters. The capabilities of the presented model are demonstrated via several fits of experimental tests on a wide range of materials. These tests involve monotonic and cyclic loading of polyurea, high-density polyethylene and brain tissue.

Application to ballistic impact on a polyurea retrofitted DH36 steel plate is simulated and validated utilizing the presented soft material model for the polymer and a porous plasticity model for the metal. Moreover, computational capability for assessing the blast performance of metal/elastomer composite shells is also presented. 

Another implemented application is in the area of traumatic brain injuries under impact/acceleration loading. Clinically observed brain damage is reproduced and a predictive capability of the distribution, intensity and reversibility/irreversibility of brain tissue damage is demonstrated.

November 27, 2007 (Thesis Seminar)
Michael Epstein
Managing Information in Networked and Multi-agent Control Systems
In recent years the field of Networked Control Systems (NCS) has emerged to describe situations where information in feedback control loops is passed through imperfect communication channels that can result in quantized, delayed and even lost information. 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 transmtited to a remote plant via a lossy communication link with no acknowledgment 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 intellligently manage the information flow and provide analysis and simulations to illustrate their effectiveness.

November 29, 2007 (Thesis Seminar)
Eric Johnsen
Numerical Simulations of Non-spherical Bubble Collapse with Applications to Shockwave Lithotripsy
Shockwave lithotripsy (SWL) is an extracorporeal procedure in which shockwaves are focused on kidney stones in an attempt to break them.

Because the stones are usually immersed in liquid, cavitation occurs and contributes to stone comminution. In the present work, numerical simulations are employed to study the collapse of a single bubble in SWL.

In order to capture interface deformations and shockwave propagation, the Euler equations, closed by a stiffened equation of state and augmented by appropriate advection equations, are solved. A new high-order accurate, quasi-conservative, shock- and interface-capturing scheme is developed to simulate bubble collapse problems. Shock-induced collapse and Rayleigh collapse in a free-field and near a wall are simulated; because of asymmetries in the flow field, the collapse is no longer spherical and a re-entrant jet forms. The general properties of non-spherical collapse are investigated. In particular, the mechanism of jet formation and the sequence of shockwaves emitted during collapse are studied. In addition, as an indication of potential damage, the pressure along a nearby solid surface is measured. The analysis is extended to SWL conditions, in order to determine the impact of bubble collapse on stone comminution.

February 13, 2008 (Thesis Seminar)
Jiang Hao
301 Thomas
Adaptive Feature Selection in Pattern Recognition and Ultra-wideband Radar Signal Analysis
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.

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. E xperimental 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.

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 T² 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.

April 16, 2008 (Thesis Seminar)
Ebraheem Fontaine, Caltech
306 Thomas
Automated Visual Tracking for Behavioral Analysis of Biological Model Organisms
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. 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. 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.

April 17, 2008 (Thesis Seminar)
John M. Carson, III, Caltech
4:00 p.m.
Robust Model Predictive Control with a Reactive Safety Mode
Control policies designed for practical engineering applications, such as aerospace and mechanical vehicles, must provide adherence to physical state and control constraints and be robust to uncertainty affecting the system dynamics and constraints. When the algorithms that produce these policies are pushed online (e.g., policies generated by using onboard computers), the algorithms must be computationally efficient and reliable. The contributions in this thesis build upon the framework of MPC (Model Predictive Control) to create a computationally-efficient, robust MPC algorithm with guaranteed re-solvability and a reactive safety mode, available at any time, to ensure system safety from changes in state constraints (e.g., other vehicles crossing/stopping in the feasible path, or unexpected ground proximity in spacecraft landing scenarios).

The framework of MPC, also known as receding horizon control, makes use of a nominal dynamics model to predict and optimize system response to a feedforward control policy that is computed online by recursively re-solving a finite-horizon optimization problem.

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. A robust MPC algorithm with guaranteed re-solvability is developed by adding a separate feedback policy to generate an invariant tube to ensure actual system trajectories remain in the proximity of the feedforward nominal trajectory at all times without violating state or control constraints. The tube is constructed through a characterization of the uncertainty between the nominal model and the actual system dynamics. To address uncertainty in state constraints, a reactive safety mode is blended into the control algorithm. The safety mode, if activated, guarantees containment within an invariant set about a safety reference for all time and guarantees satisfaction of control and safety state constraints.

April 22, 2008
Jacopo Buongiorno, MIT
Heat Transfer Enhancement in Nanofluids
Colloidal dispersions of nanoparticles are known as 'nanofluids'. Such engineered fluids offer the potential for enhancing transport phenomena, particularly heat transfer, while avoiding the drawbacks (i.e., erosion, settling, clogging) that hindered the use of particle-laden fluids in the past. At MIT we have been studying the heat transfer characteristics of nanofluids for the past 3½ years, with the goal of evaluating their benefits for and applicability to conventional and nuclear power systems. This presentation will survey the MIT research in this area with particular emphasis to nanofluid thermo-physical property determination, single-phase heat transfer measurements and interpretation, and boiling behavior, including, prominently, the Critical Heat Flux limit.

April 25, 2008 (Thesis Seminar)
Angel Ruiz Angulo, Caltech
2:00 pm
Surface Deformation in a Liquid Environment Resulting from Single Particle Collisions
Multiphase flows are fairly complex and they are usually studied as a bulk. The way we approach this problem is by looking at single particle interactions (particle-particle and particle-wall). This thesis presents experimental measurements of the approach and rebound of a particle colliding with a "deformable" surface in a viscous liquid. The complex interaction between the fluid and the solid phases is coupled through the dynamics of the flow as well as the deformation process. Steel particles were used to impact different aluminum alloy samples (Al−6061, Al−2024, and Al−7075) using different aqueous mixtures of glycerol and water as a viscous fluid. Normal coefficient of restitution and deformation parameters account for losses due to lubrication effect and inelasticity, identifying then, the dominant energy loss mechanism during the collision process. The experiments clearly show four different regimes depending on the impact Stokes number: lubrication effect and elastic deformation, lubrication effect and elastic-plastic deformation, elastic deformation with no hydrodynamic effects, and merely elastic-plastic deformation with negligible lubrication effect. An analysis of the erosion of ductile materials during immersed collisions is presented. The size of the crater formed by the impact of a single particle against a ductile target can be estimated from theory, and these estimates agree well with experimental measurements.

April 29, 2008
Denis J. Phares, USC
Real-time Chemical Analysis of Aerosols: An Approach to Identifying Organic Compounds
Understanding how small particles in the atmosphere affect health and the environment requires knowledge of their chemical composition. Issues associated with bulk aerosol analysis, such as low temporal resolution, size biases, and chemical transformation after sampling, has led to the development of aerosol mass spectrometers that can determine the size and chemical composition of ambient aerosols in real-time. Some of these instruments have provided quantitative data concerning the content of various salts and metals present in the aerosol. However, identification of organic compounds is more difficult because of fragmentation that occurs during vaporization and ionization, and because of the complicated mass spectra that are generated from particles that contain mixtures of organics. This talk will focus on new developments in instrumentation aimed at addressing some of these issues.

May 13, 2008
Wendy Zhang, University of Chicago
Memory-encoding Shape Vibrations in a Disconnecting Air Bubble
Recent experiments have discovered that how an underwater air bubble disconnects is exquisitely sensitive to slight asymmetries in its neck shape. Here I show that the classic approach of modeling the disconnection as a 2D Rayleigh-Plesset collapse provides a simple explanation for this sensitivity. The cylindrically-symmetric inviscid collapse is Hamiltonian, thus naturally possessing a complete memory of the energy distribution initiating the disconnection. A linear stability analysis reveals the singularity dynamics controlling the final moments of the disconnection also has a precise memory of the azimuthal energy distribution. This memory is encoded by constant-amplitude vibrations in the cross-section shape of the bubble neck.  Finally I describe an effort to directly check the relevance of this mechanism. In an experiment by Keim and Nagel, a mode-2 vibration about the cylindrically-symmetric disconnection dynamics is excited by causing an air bubble to detach from a slot-shaped nozzle. A comparison finds good agreement between the measured vibration dynamics and the calculated dynamics.

May 13, 2008
Sam Taira, Caltech (Thesis Seminar)
1:00 p.m.
The Immersed Boundary Projection Method and its Application to Simulation and Control of Flows around Low-Aspect-Ratio Wings
We introduce a new formulation of the immersed boundary method capable of simulating incompressible flows over arbitrarily shaped bodies in motion and/or under deformation. The no-slip condition along the immersed boundary and the incompressibility constraint are enforced simultaneously through a single projection in the present approach, referred to as the immersed boundary projection method. The boundary force is determined implicitly without any constitutive relations for the rigid body formulation, allowing the use of high CFL numbers in our simulations compared to past methods.

The current method is used to analyze three-dimensional flows around purely translating low-aspect-ratio flat-plate wings. Simulations highlighting the unsteady nature of the separated flows are performed at low Reynolds numbers for various aspect ratios, angles of attack, and planform geometries. We observe that the stability of the wake and the force experienced by the wing are mostly influenced by the aspect ratio and the angles of attack. At high angles of attack, the leading-edge and tip vortices interact strongly as they detach from the airfoil. Since the detachment of leading-edge vortices results in the loss of added lift, flow control with steady blowing is considered for modifying the wake structure in a favorable manner . With a suitable actuator setup, the tip vortices are strengthened by engulfing the trailing-edge vortex sheet, which increase the downward thrust and the downward-induced velocity onto the leading-edge vortices. The tip vortices are therefore used to increase lift in post-stall flows for the considered low-aspect-ratio wings.

May 16, 2008
Michael Wolf, Caltech (Thesis Seminar)
306 Thomas
Target Tracking Using Clustered Measurements, with Applications to Autonomous Brain-Machine Interfaces
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. 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".

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. 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.

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.

May 20, 2008
Eric Lauga, UCSD
Nonlinear and Nonlocal Hydrodynamics of Swimming Microorganisms
Microorganisms swimming in viscous fluids inhabit a world quite different from the one we are used to experiencing. In this talk, we will discuss some properties and recent results of fluid-based locomotion on very small scales in fluids and geometries where nonlinear behavior naturally arises. We will first discuss experimental and theoretical results on the hydrodynamic attraction of swimming cells by solid surfaces. We will then present theoretical work on locomotion in viscoelastic fluids.

May 21, 2008
Mary Dunlop, Caltech (Thesis Seminar)
Dynamics and Correlated Noise in Gene Regulation
Gene regulatory interactions are context dependent, active in some cell types or cellular states but not in others. I will present a method for determining when a regulatory link is active given temporal measurements of gene expression. Correlations in time series data are used to determine how genes influence each other and their causal relationships.

Natural stochastic noise is shown to aid in the process of network identification by perturbing the expression of genes; the speed and direction at which the noisy signal propagates shows how the network is connected.

May 27, 2008
Winston Jackson, Caltech (Thesis Seminar)
10:30 a.m., 306 Firestone
Characterization of Soft Polymers and Gels Using the Pressure-Bulge Technique
A method to characterize the bulk hydrated properties of soft polymers and hydrogels, whose moduli are in the low MPa regime, using the pressure-bulge technique is presented. The pressure-bulge technique has been used extensively in the characterization of thin film behavior, particularly for the case of metals. The extension of the plane-strain and circular bulge techniques to determine the Young's modulus and Poisson's ratio of bulk latex and silicone rubber sheets are shown here, in addition to the viscoelastic behavior of 5% agarose gel in the time domain using relaxation tests.

The membranes are clamped between two stainless steel plates that are connected to a liquid pressure chamber. A syringe connected to a linear actuator causes changes in the pressure and displacement, and a pressure sensor and confocal displacement sensor are used to monitor these changes in real time. The theory presented converts the measured pressure and displacement data into stress and strain data, using a geometrically nonlinear analysis, and the elastic and viscoelastic modulus properties are then determined from fits to this data.

The results from the bulge test experiments are compared with data from hydrated uniaxial tension tests, and the data comparison with respect to the all of the materials tested show good agreement between the two tests. These results show promise regarding the use of pressure-displacement techniques to characterize other soft material systems, including biological polymers and tissues as well as cell-to-matrix and cell-to-cell interactions under varying mechanical loading conditions of cell substrates.

May 27, 2008
Shannon Kao, Caltech (Thesis Seminar)
Detonation Stability with Reversible Kinetics
Detonation propagation is unsteady due to the innate instability of the reaction zone structure. Up until the present, investigations of detonation stability have been exclusively concerned with model systems using the perfect gas equation of state and primarily single-step irreversible reaction mechanisms.

This study investigates detonation stability characteristics with reversible chemical kinetics models. To allow for more general kinetics models, we generalize the perfect gas, one-step irreversible kinetics, linear stability equations to a set of equations using the ideal gas equation of state and a general reaction scheme. We linearly perturb the reactive Euler equations following the method of Lee and Stewart (1990) and Short and Stewart (1998). Our implementation uses Cantera (Goodwin, 2005) to evaluate all thermodynamic quantities and evaluate generalized analytic derivatives of quantities dependent on the kinetics model.

The computational domain is the reaction zone in the shock-fixed frame such that the left boundary conditions are the perturbed shock jump conditions which we have derived for a general equation of state and implemented for an ideal gas equation of state. At the right boundary, the system must satisfy a radiation condition requiring that all waves travel out of the domain. Unlike the case of a single reversible reaction, in a truly multistep kinetics model, the radiation boundary condition cannot be solved analytically. In this work, we provide a general methodology for satisfying the appropriate boundary condition.

We then investigate the effects of reversibility on the characteristics of the instability in one and two dimensions. These characteristics are quantifed by the unstable eigen values as well as the shape of the base flow and eigenfunctions. We show that there is an exchange of stability as a function of reversibility. To validate our work, we have performed unsteady calculations. We show that we can match the frequencies predicted by our linear stability calculations near the stability threshold.

September 11, 2008
Ling Zheng, Caltech (Thesis Seminar)
Wrinkling of Dielectric Elastomer Membranes
10:00 a.m.

Wrinkling of thin membranes due to different in-plane loading and boundary conditions has drawn attention of researchers in structural engineering since the development of thin webs for early aircraft structures. More recently, pre-stressed lightweight membrane structures have been proposed for future space missions, for example solar sails, the next generation space telescope sunshield and space-based radar systems. These structures are often partially wrinkled during operation. The formation of wrinkles alters the load paths and the structural stiffness of the membranes. More importantly its occurrence degrades the surface accuracy of these structures, which is a key design parameter.

This dissertation focuses on wrinkling of thin rectangular membranes subjected to uniaxial tension and investigates the onset and profiles of wrinkles using both experimental and numerical approaches.

An optical method, which integrates fringe projection method with four-frame phase-shifting technique, pre-conditioned conjugate gradient phase unwrapping algorithm and series-expansion carrier removal technique was developed in order to measure the full-field out-of-plane displacement of membranes, and an optical system was constructed including a uniaxial tension testbed , a LCD projector and a CCD camera. A series of uniaxial tensile tests were carried out on silicone rubber membranes of varying dimensions and aspect ratios in order to investigate the effect of geometric factors such as membrane dimension and aspect ratio on wrinkling onset; and a series of measurements were performed on each membrane at several desired strain levels to understand the evolution of the wrinkles, in particular wrinkle amplitude and wavelength.

A numerical study was carried out using the commercial finite element software ABAQUS to further understand the important characteristics of wrinkling of thin   membranes observed in thee physical model. Geometrically nonlinear finite element models of membrane structures were constructed with thin-shell elements. A series of simulations were carried out for different membrane dimensions. The critical buckling load and buckling modes was predicted for each dimension using a pre-buckling eigenvalue analysis. The desirable buckling mode was selected and introduced into the structure as a geometric imperfection. The formation and growth of wrinkles were simulated in the post-buckling analysis.

Finally, an idea of suppressing wrinkle instabilities of dielectric elastomer membranes using through-thickness electric field was proposed and verified in both experiment and numerical simulations.

Keywords: wrinkling; membrane; critical buckling strain; wrinkle onset; wrinkle wavelength; wrinkle amplitude; fringe projection method; finite element; thin shell;  dielectric elastomer; dielectric actuation.

September 22, 2008
Nick Hudson, Caltech (Thesis Seminar)
10:00 a.m.
Inference in Hybrid Systems with Applications in Neural Prosthetics

New methods for the identification of several classes of hybrid systems that are characterized by discrete switching between sets of continuous dynamical activity are discussed. Inspired by the characterization of biological systems into discrete sets of behaviors or modes, the developed models use a mixture of duration, time and dynamical state based switching paradigms. Specifically, both stationary and non-stationary Markov chains are used to govern mode switching, while generalized linear models are used to represent the set of continuous dynamics. To identify hybrid system models from data, a Bayesian framework is used, so as to facilitate incorporation of prior knowledge in a coherent way, and provide a basis for selection between a set of possible models. The inherent difficulty in identifying hybrid or switching systems from data is a consequence of the discrete system states being `hidden', and not observed. Thus, in identifying this class of systems, simultaneous identification of the continuous dynamics and classification of the observed data into discrete model states is required. To approach the identification problem in a structured way, the variational Bayesian framework, the Gibbs sampler and the expectation maximization algorithm are adapted for inference in developed models.

The developed methods are used for model generation in cortical neuroprosthetic medical devices that aim to help the severely handicapped. In such systems, a ``supervisory decoder'' is required to classify the activity of extracellular neural recordings into a discrete set of modes that model the evolution of the brain's planning process. In moving from experimental, laboratory based prosthetic development programs, to clinical applications and rehabilitation of patients, new automated methods for generating supervisory decoders and control systems are required.  The proposed supervisory decoding framework is demonstrated on recorded neural data during a delayed center-out reach experiment.