Distributed Manipulation
Todd Murphey, Joel W. Burdick

This research analyzes the stability of distributed manipulation control schemes. A commonly proposed method for designing a distributed actuator array control scheme assumes that the system's control action can be approximated by a continuous vector force field. The continuous control vector field idealization must then be adapted to the physical actuator array. However, we have shown that when one takes into account the discreteness of actuator arrays and realistic models of the actuator/object contact mechanics, the controls designed by the continuous approximation approach can be unstable. For this analysis we introduce and use a ``power dissipation'' method that captures the contact mechanics in a general but tractable way. We show that the quasi-static contact equations have the form of a multi-model hybrid system. We introduce a discontinuous feedback law can produce stability which is robust with respect to variations in contact state. (full report)

A Priori Training Data Valuation
Alexander Nicholson, Yaser Abu-Mostafa

For machine learning it is generally accepted that a greater amount of available data facilitates improved generalization. In practice, however, a learning algorithm cannot accomodate and unlimited data set and may be hindered by noisy and irregular data. We introduce a procedure for evaluating individual training examples. This valuation can serve as a basis for selecting training sets of limited size and for detecting outliers or other undesirable data. We demonstrate that learning with a data set from which the worst data has been removed can result in improved generalization performance. (full report)

The Bin Model for Generalization
Alexander Nicholson, Xubo Song, Yaser Abu-Mostafa

The problem of overfitting the data is attacked by using the Bin Model analysis. This provides a method of bounding generalization error without sacrificing valuable training data. (full report)

Gesture Recognition
George Panotopoulos, Dinkar Gupta, Demetri Psaltis, Pietro Perona

Though your personal computer has a processing capacity orders of magnitude larger than it did some ten years ago you still use the same means to interface with it, namely a keyboard and pointing device. In the context of this project we investigate the design of an interface based on human gestures. The system we are envisioning is not limited to a particular user and should be able to learn new gestures.
(full report)

Models of Visual Object Categorization In Humans
Robert J. Peters, Fabrizio Gabbiani, Christof Koch

Previous studies of exemplar, prototype, and decision-bound models of visual object categorization have not resolved the importance of memory capacity and flexibility of decision surfaces in human categorization behavior. We have compared these previous models with our new roaming exemplar model (RXM), according to their abilities to match human observers' categorizations of various 2-D image contours. Unlike past comparisons among categorization models, we explicitly accounted for memory capacity by penalizing models for their number of free parameters with the Akaike information criterion. This revealed that a successful model of human categorization--such as the RXM--did not require a large memory capacity if the orientation of its decision boundary was unconstrained, suggesting that an efficient computer implementation of object categorization could also rely on limited memory storage. (full report)

Flexible Parylene-Valved Skin for Adaptive Flow Control
T. Nick Pornsin-Sirirak, Matthieu Liger, Yu-Chong Tai, Steve Ho, Chih-Ming Ho

This research describes the first work of using wafer-sized flexible parylene-valved actuator skin (total thickness ~ 20 _m) for micro adaptive flow control. The check-valved actuator skin features vent-through holes with tethered valve caps on the membrane to regulate pressure distribution across the skin. The skins were integrated onto micro-aerial-vehicle (MAV) wings that were tested in the wind tunnel for aerodynamic evaluation. The test result has shown a very significant effect on the aerodynamic performance. Compare to the reference wings (no actuators), both the lift and thrust of the check-valved wings are improved by more than 50%. This is the first experimental result to demonstrate that the application of MEMS actuator skins for flow control is very promising. (full report)

Structural Description of Basic Objects With Features
Christoph Rasche

We explore the representation of basic-level categories using computer vision methods. The category representation is expressed by lines, arcs and a combination thereof. In a bottom-up process we extract such features, in a top-down process we try to match each category representation against the bottom-up output. (full report)

Attention Modulation of Visually Responsive Neurons in the Human Medical Temporal Lobe.
Leila Reddy, Patrick Wilken, Christof Koch

Previous work from our laboratory (Kreiman et al.,2000) has shown that neurons in the medical temporal lobe structures are visually responsive to categories of images. We intend to test whether attention modulates cell firing in these neurons.
(full report)

Face-gender Discrimination Modulated by Attentional Load
Leila Reddy, Patrick Wilken, Christof Koch

This experiment demonstrates that performance in gender discrimination tasks is compromised when attention is engaged by another attentionally demanding task. However, performance is still highly above chance implying that in the near absence of attention, observers can still distinguish the gender of a face to some extent. (full report)

Shadow Carving
Silvio Savarese, Holly Rushmeier, Fausto Bernardini, Pietro Perona

The shape of an object may be estimated by observing the shadows on its surface. Assuming that a conservative estimate of the object shape is available, our method analyzes images of the object illuminated with known point light sources and taken from known camera locations. The surface estimate is adjusted using the shadow regions to produce a refinement that is still a conservative estimate. A proof of correctness is provided. The method has been tested and validated with experimental results. (full report)

Detection of Human Motion in a Cluttered Scene
Yang Song, Luis Goncalves, and Pietro Perona

Humans are the most important component of a machine's environment. We develop an algorithm which can generate models of human motion automatically from unlabeled real image sequences. Experiments show that the resulting models can successfully detect and label humans from image sequences with clutter and occlusion. (full report)

Attentional Modulation of Visual Motion Perception Using Novel Wavelet Stimuli Ð Combination Study of Psychophysics and fMRI Imaging
N. Tsuchiya, G. Rees, J. Braun & C. Koch

We have previously characterized the effects of withdrawing attention on detection and discrimination of static visual stimuli (Lee et al. Nat Neuro 1998). Here we report attentional modulation of motion perception in psychophysics experiment. A novel motion stimulus comprising spatio-temporally contrast-modulated Gabor wavelets was used to distinguish attentional effects on mechanisms sensitive to component motion from those sensitive to pattern motion (Schrater et al Nat Neuro 2000). In the second experiment, we confirmed our component stimulus only activates only early visual cortex by functioal magnetic resonance imaging (fMRI) measurement, supporting our argument in psychophysics and consistent with our previous result (Rees et al. Nat Neuro 2000). (full report)

Part 1/ Rapid Visual Categorization In The Absence of Awareness
Part 2/ Processing Capacity For Natural Scenes and Objects in the Human Visual System
Rufin VanRullen

Humans can categorize natural scenes on the basis of the presence of a target object (i.e. animal) so rapidly (150 ms) that such processing has been proposed to rely on the feed-forward propagation of information collected during the first milliseconds of visual stimulation. According to this view, early motor responses should be mostly unaffected by masking the visual stimulus after a few tens of milliseconds. We asked our subjects to respond to masked (SOA 26.6 ms) and unmasked natural scenes when they contained an animal. (full report)

Primitives for Human Motion: A Dynamical Approach
D. Del Vecchio, R.M. Murray, P. Perona

Using tools from d dynamical systems theory and systems identification theory we develop the study of primitives for human motion which we refer to as movemes. We introduce basic definitions of dynamical independence of LTI systems and segmentability of signals and we develop classification and segmentation algorithms for two dimensional motions. We test our ideas on data sampled from four human subjects who were engaged in a simple real­life activity including two movemes. Our experiments show that we are able to distinguish between the two movemes and recognize them even when they take place in an activity containing more than one moveme.
(full Report)

Towards an Integrated Model of Saliency-based Attention and Object Recognition
Dirk Walther, Maximilian Riesenhuber, Tomaso Poggio, Laurent Itti, Christof Koch

We are working on an integrated model for the dorsal (where) and the ventral (what) pathway in the primate's visual processing system and the interaction between these two pathways. The model will be applied to visual search tasks for detecting objects in cluttered natural scenes. Components of top-down attention will be integrated into the system to achieve this goal. (full report)

Underwater Shear Stress Sensor
Yong Xu, Fukang Jiang, Qiao Lin, Jason Clendenen, Steve Tung and Yu-Chong Tai

A micromachined, vacuum-cavity insulated, thermal shear stress sensor is developed for underwater applications. The two major challenges for underwater application, namely the waterproof coating and pressure sensitivity, are specially studied for our device. (full report)

Swarm Intelligence and Traffic Safety
Yizhen Zhang, Alcherio Martinoli
Collaborators: Erik Antonsson (Caltech), Ross Olney (Delphi-Delco Automotive Systems)

A smart car that assists the driver must give warnings in dangerous situations, override the driver to avoid collisions, and help to reach the intended destination as quickly as possible. Unfortunately, satisfying these requirements and at the same time leaving the decisional autonomy at the individual level becomes an extremely hard problem to solve with traditional methods. Biologically-inspired techniques such as Swarm Intelligence and Incremental Evolution provide new promising ways to tackle the design and distributed control problems of a traffic system. In this project, solutions are developed using embodied simulations and validated with real robot experiments. (full report)