ijcai-99 Home Page

B3 - Behavior-based Robotics

Monday, AM

Maja Mataric &Ronald Arkin

Behavior-based robotics has, in the last decade, emerged as one of the leading approaches to mobile robot control and has been effectively applied in a variety of domains, ranging from modeling biological systems, to studying difficult robotics problems, to real-world applications. Behavior-based control addresses the fundamental AI issues of sensing, thinking and acting in realtime and presents a successful approach to solving situated AI problems.

In this tutorial we present a brief history of intelligent robotics, describe the interdisciplinary origins of behavior-based control, and place it in historical context relative to classical deliberative approaches, reactive control, and the currently most popular hybrid control. We illustrate the basic principles of behavior-based control, and methods for system synthesis and analysis. We present an overview of relevant biological inspirations and models of robot control, from a neuroscientific, ethological, and psychological perspective. Key issues in perception for behavior-based systems, including active, action-oriented, and modular perception, are covered. We conclude by surveying the current state-of-the-art in research and applied control, and outline the outstanding problems and current directions, including robot learning and multi-robot control. Numerous videotapes of robots in action are used to illustrate and evaluate the presented concepts.

The target audience is the general AI community; the tutorial ties behavior-based robotics to general AI methods, principles and goals. It gives a clear idea of the current state of a major area in robotics, thus making many of the talks in the robotics sessions, as well as the IJCAI and AAAI Robotics contest and Exhibition demonstrations, more interesting and accessible.

Prerequisite knowledge:
The tutorial will not require any robotics background on the part of the audience. A general AI background (at the level of an undergraduate AI course) will be sufficient to follow all of the material in the tutorial.

Maja Mataric is an assistant professor in the Computer Science Department and the Neuroscience Program at the University of Southern California, and Director of the USC Robotics Research Labs. She received her PhD in Computer Science and AI in 1994 and her MS in 1990, both from MIT. She is on the editorial board of JAIR and Adaptive Behavior. Her research interests include multi-robot and multi-agent control and learning, and modeling imitation. She is a member of AAAI and ISAB.

Ronald Arkin is a Professor in the College of Computing at Georgia Tech and is Director of the Mobile Robot Laboratory. His interests include behavior-based reactive control and action-oriented perception for mobile robots and unmanned aerial vehicles, robot survivability, multi-agent robotic systems, and learning in autonomous systems. He recently completed a textbook entitled "Behavior-Based Robotics" (MIT Press) and is the Series Editor for the book series Intelligent Robotics and Autonomous Agents. He is a Senior Member of the IEEE, and a member of AAAI and ACM.

Webmaster: Sven Olofsson, sveno@dsv.su.se
Last modified: Mar 16, 1999