Department of Electrical and Computer Engineering

Convergence to Nash Equilibrium in Multi-agent Decision Systems, Dr. Reza Olfati-Saber, PhD. MIT

Scalable Coordination of Networked Systems and
Collaborative Sensor Fusion
 
Dr. Reza Olfati-Saber
PhD. MIT

Time: 11-12, March 04, 2005,  (Friday)
Place: Coover 2222

Abstract: Networked dynamic systems have broad industrial, commercial, and military applications ranging from groups of unmanned autonomous vehicles (UAVs) performing cooperative surveillance to sensor networks, genetic & biological networks, and smart materials with embedded sensors & actuators. Coordination of such large-scale systems with many interacting subsystems presents new challenges in terms of control, scalability, robustness, and performance issues.  The first part of the talk discusses our recent work on Consensus Problems for multi-agent dynamic systems and how they enable the derivation of distributed Kalman filters for multi-sensor fusion and collaborative information processing. In the second part, a scalable algorithm is presented for collective control of Swarms of mobile agents. This swarming algorithm is derived analytically and has guaranteed convergence properties. Swarms can act as mobile sensor networks. A multi-species particle system is introduced to model the motion of swarms in presence of multiple obstacles.  Swarms are the focus of attention of numerous scientists from diverse backgrounds in systems and controls, computer science, statistical physics, biophysics, and mathematical biology as well as several industry and government agencies.
The third part of the talk focuses on performance issues in complex networked systems. It is illustrated that how Small-World Networks play a crucial role in improving the speed of consensus algorithms by multiple orders of magnitude. This is due to a novel spectral phase transition phenomenon. Several examples and simulation results are provided to illustrate the effectiveness of the proposed analytical and computational tools.

Bio: Dr. Olfati-Saber received his S.M. and Ph.D. degrees in EECS from MIT in 1997 and 2001, respectively, and his B.S. degree (with honors) in EE from Sharif University of Technology in 1994. From 2001 to 2004, he was a postdoctoral scholar at the department of Control & Dynamical Systems at Caltech. He is currently a visiting scientist at the department of Mechanical & Aerospace Engineering at UCLA. He is an elected member of Sigma Xi scientific society and a member of IEEE and AIAA. He won the Lotfi Zadeh best student paper award of the NAFIPS ?96 conference at UC, Berkeley.