CS Colloquium
Thursday, October 29, 2009
4:00 p.m., AKW 200

Sign up to meet with speaker.

Host: Brian Scassellati

Speaker: Chad Jenkins, Brown University
Title: Manifold Learning in Human-Robot Teams

Abstract: A principal goal of robotics is to realize embodied systems that are effective collaborators in human endeavors pursued in the physical world. Human-robot collaborations can occur in a variety of forms, including autonomous robotic assistants, mixed-initiative robot explorers, and augmentations of the human body. For these collaborations to be effective, human users must have the ability to realize their intended behavior into actual robot control policies. At run-time, robots should be able to manipulate an environment and engage in two-way communication in a manner suitable to their human users. Further, the tools for programming, communicating with, and manipulating using robots should be accessible to the diverse sets of technical abilities present in society. Towards the goal of effective human-robot collaboration, learning from demonstration (LfD) has emerged as a central theme of our work for the natural instruction of autonomous robots by human users. In robot LfD, desired cognitive functions for a robot (perception, decision making, or motion control) are implicit in human demonstration rather than explicitly coded in a computer program.

In this talk, I will present our work into learning priors from human demonstration for robot perception and control using manifold-based dimension reduction. My specific focus will be the development and application of manifold learning algorithms to estimate subspace priors for spatial and time-series data generated by humans. I will discuss our approach to spatio-temporal dimension reduction in the context of manifold learning. Using manifold learning, results will be presented from learning priors for: 1) classifying tactile signatures to recognize successful grasps on the NASA Robonaut, 2) providing low-dimensional control spaces for neural prosthetics, 3) learning motion primitives from human movement data, and 4) extracting kinematic models and poses from multi-view video of human performance. Our approach to learning priors will be cast in the broader context of policy learning and computational models for communication in multi-robot multi-human systems.

Biography: Odest Chadwicke Jenkins, Ph.D., is an Assistant Professor of Computer Science at Brown University. Prof. Jenkins earned his B.S. in Computer Science and Mathematics at Alma College (1996), M.S. in Computer Science at Georgia Tech (1998), and Ph.D. in Computer Science at the University of Southern California (2003). Prof. Jenkins was selected as a Sloan Fellow and a Kavli Fellow in 2009. He is a recipient of the Presidential Early Career Award for Scientists and Engineers (PECASE) for his work in physics-based human tracking. He has also received Young Investigator awards from the Office of Naval Research (ONR) for his research in learning dynamical primitives from human motion, the Air Force Office of Scientific Research (AFOSR) for his work in manifold learning and multi-robot coordination and the National Science Foundation (NSF) for robot learning from multivalued human demonstrations. His research addresses problems in robot learning and human-robot interaction, primarily focused on robot learning from demonstration, as well as topics in computer vision, machine learning, and computer animation.