Jul 16: Seminar on Generalized Sampling based Feedback Planners

Dr. Suman Chakravorty, Associate Professor at Aerospace Engineering Department at Texas A&M University, would be presenting a seminar in the department as per the following schedule.

Speaker:   Dr. Suman Chakravorty, Associate Professor Aerospace Engineering Department at Texas A&M
Date  :      Tuesday, July 16, 2013
Time  :      04:00 PM
Venue :     AE seminar room

Abstract

In this talk, we shall consider the problem of sequential planning under process and sensing uncertainty. It is well known that such problems can be modeled using the so-called Partially Observable Markov Decision Processes (POMDP). However, it is also equally well known that such POMDPs suffer from the twin curses of "dimensionality" and "history" which results in the problem being computationally intractable for most problems residing in continuous state and control spaces. In this talk, we shall propose a sampling based hybrid, hierarchical approach to solving a class of such problems, namely stochastic shortest path problems. These planners succeed in breaking the "curse of history" for POMDPs, while the "curse of dimensionality" is greatly alleviated owing to their sampling based nature. These planners also have guaranteed performance in terms of a "probability of success". We show the application of the planners to the feedback motion planning of various holonomic and nonholonomic robotic systems, operating under both process and sensing uncertainty, including experiments on real robots, as well as experiments on multi-robot systems.

About the Speaker

Dr. Suman Chakravorty obtained his B.Tech in Mechanical Engineering in 1997 from the Indian Institute of Technology, Madras and his Doctoral degree in Aerospace Engineering from the University of Michigan, Ann Arbor in 2004. From August 2004- August 2010, he was an Assistant Professor with the Aerospace Engineering Department at Texas A&M University, College Station and since August 2010, he has been an Associate Professor in the same department. Dr. Chakravorty¹s research interests lie in the broad area of estimation and control of stochastic dynamical systems with application to autonomous, distributed robotic mapping and planning. He has served as an associate editor for the American Control Conference (ACC) and the IEEE/ RSJ International Conference on Intelligent Robots and Systems (IROS). He is a member of AIAA and IEEE.