Safety-Aware Transportation Systems: Cooperative Autonomous Driving for Vehicular Networks
Reza Olfati-Saber, Thayer School of Engineering
Friday, February 18, 2011, 3:30pm
This seminar is part of the Jones Seminars on Science, Technology, and Society series
The existing transportation systems are mostly controlled by humans who are prone to making errors causing collisions, fatalities, and unnecessary traffic congestion. High levels of road congestion significantly contributes to air pollution in large cities, wasting fuel, and adding to the frustration of drivers. More than a million road accident fatalities occur world-wide every year. Some of the main causes of traffic accidents are 1) weather and road conditions; 2) drivers and pedestrians being distracted or intoxicated; and 3) the inability of human drivers to predict and react to imminent threats. Our objective is to design the first generation of "safety-aware transportation systems" in which the vehicles are equipped with embedded sensing, control, computing, and communication devices. The embedded sensors enable the vehicles to detect and track the roads, dividers, and the surrounding pedestrians, bikers, and vehicles (i.e., obstacles). In this talk, we demonstrate that driving on a lane is a nontrivial task for a fleet of vehicles requiring beyond pairwise inter-vehicle interactions. Furthermore, a safe way of passing slower vehicles requires a detailed set of considerations and actions similar to manual driving by humans. Our approach to design of autonomous driving algorithms is more inspired by flocking behavior of birds during migrations than human behavior. Using tools from distributed control theory for networked multi-agent systems and nonlinear control, we design novel driving algorithms that enable vehicular networks to autonomously perform tasks such as lane-driving, lane-change, passing, handling intersections, and braking. We present both experimental and analytical results that illustrate the effectiveness of our cooperative autonomous driving algorithms.
About the Speaker
Reza Olfati-Saber received his PhD and SM degrees from Massachusetts Institute of Technology in 2001 and 1997, respectively, in Electrical Engineering and Computer Science. He was a postdoctoral scholar at California Institute of Technology (2001-04) and a visiting scientist at UCLA (2004-05) prior to joining Dartmouth. He is currently an assistant professor of engineering at the Thayer School of Engineering. Dr. Olfati-Saber is the recipient of the 2010 PECASE (Presidential Early Career Award for Scientists and Engineers) and the 2008 NSF CAREER award. He is the author of two seminal papers on "consensus problems" and "flocking". His research interests include distributed control and estimation, robotics, cyber-physical networked systems, and behavioral modeling.