Broad Area Colloquium For AI-Geometry-Graphics-Robotics-Vision


Principled Methods for Behavior-Based Control and Learning Applied to Robot Teams and Humanoids

Maja J Mataric'
University of Southern California

Wednesday, May 24, 2000
refreshments 4:05PM
talk begins 4:15PM
TCseq201, Lecture Hall B
http://robotics.stanford.edu/ba-colloquium/

Abstract

Behavior-based control, which exploits the dynamics of collections of concurrent, interacting processes coupled to the external world, is both biologically relevant and effective for control problems featuring local information, uncertainty, and non-stationarity. In this talk we describe methods we have developed for principled behavior-based control and learning in two problem domains: multi-robot coordination and humanoid imitation.

In the multi-robot domain the key challenges involve reconciling individual and group-level goals and achieving scalable, on-line real-time learning. How to do all of this in a distributed behavior-based way in a timely and consistent fashion? We describe methods for Pareto-optimal behavior selection for principled group coordination, publish/subscribe messaging for distributed communication, and augmented Markov models for on-line real-time model building for group adaptation. The results are demonstrated on groups of locally-controlled but globally efficient mobile robots performing distributed collection, multiple-target-tracking, and object manipulation. In the second part of the talk we describe the application of behavior-based control in the form of basis behaviors or primitives to the problem of humanoid control. Here, the challenges include the high dimensionality of the system and the need for tight coupling between the perceptual and motor systems. We describe an imitation model that employs direct sensory-motor mappings within the behavior-based framework to segment and map the observed movement onto the existing motor system. The same biologically motivated method facilitates recognition, classification, prediction, and learning. The results are demonstrated on a 20 degree-of-freedom dynamic humanoid imitating dance and sports movements from visual data of human demonstrations.

About the Speaker

Maja Mataric is an assistant professor in the Computer Science Department and the Neuroscience Program at the University of Southern California, Director of the USC Robotics Research Labs and Associate Director of the Institute for Robotics and Intelligent Systems. She received her PhD and MS degrees in Computer Science and AI from MIT in 1994, and 1990, respectively. She is a recipient of the NSF Career Award, the MIT TR100 Innovation Award, the IEEE Robotics and Automation Society Early Career Award, and is featured in the upcoming movie about scientists, "Me & Isaac Newton". In her collaborations, she has interacted with a variety of human and robotic colleagues (ranging from LEGO robots to humanoids) at NASA's Jet Propulsion Lab, the Free University of Brussels AI Lab, LEGO Cambridge Research Labs, GTE Research Labs, the Swedish Institute of Computer Science, and ATR Human Information Processing Labs in Japan. Her Interaction Lab at USC performs research in the areas of control and learning in behavior-based multi-robot systems and humanoids.
bac-coordinators@ cs.stanford.edu
Back to the Colloquium Page
Last modified: Mon May 1 11:58:48 2000