             |
GRASP Seminar Series: Fall 2006
November 10, 11:00 a.m., Wu & Chen Auditorium
Chad Jenkins
Brown University
Abstract: There exists a divide between the needs of the general population and the capabilities of current autonomous robotic systems. In order to meet these needs and achieve greater ubiquity, general users of robot technology must be able to form and tailor control policies that reflect the intended robot behavior. While robots are increasingly able to perceive and actuate in various environments, the interface between humans and robots often remains limited to a remote control and/or computer programming. Such interfaces inhibit a user's ability to express and realize their intended control policy onto an autonomous robot as they demand constant human supervision or technical expertise.
Toward bridging this divide, our research has actively pursued robot learning from demonstration (LfD) as one means for human users to naturally instruct autonomous robots. In robot LfD, the desired robot control policy is implicit in human demonstration rather than explicitly coded in a computer program. The robot must be able to: 1) perceive relevant aspects of the demonstration, 2) infer a control policy that accords with the demonstration, and 3) receive user feedback for refinement of the inferred policy.
In this talk, I will present our work towards realizing robot LfD in various robot domains through the application and development of new methods in learning, vision, and physical control. In particular, this presentation will discuss work related to unsupervised learning of dynamical human motion primitives, markerless and untethered motion capture, balance and control of physically simulated humanoids, uncovering sensorimotor manifolds in robot manipulation, and learning robot behavior fusion.
Biography: Odest Chadwicke Jenkins,
Ph.D., is Assistant Professor of Computer Science at Brown University.
He earned his Ph.D. in Computer Science at the University of Southern
California (2003). Prof. Jenkins' research interests focus primarily
on robot learning, human-robot interaction, and motion capture systems.
His broader interests include humanoid and mobile robotics, machine
learning, computer vision, computer animation, interactive entertainment,
and neural prosthetics.
GRASP
Laboratory
Site maintained by graspadm@grasp.cis.upenn.edu
Last update:17 October, 2006