School of
Information Technology and Electrical Engineering

Towards a Principled General Approach to Robot Motion Planning under Uncertainty

Prof David Hsu - National University of SingaporeMon, 16/07/2012 - 14:00
Dr Surya Singh

A major challenge in robot motion planning is the uncertainty inherent in robot control and sensing.  The partially observable Markov decision processes (POMDPs) provides a powerful framework to incorporate uncertainty into planning and greatly improve the reliability of robot operation.  To deal with imperfect state information, POMDP planning reasons with beliefs, i.e., probability distributions over possible robot states.  Although POMDP planning is computationally intractable in the worst case, there has been rapid progress recently in developing efficient approximation algorithms.  Today, moderately complex POMDPs with 100,000s states can be solved in reasonable time.  I will present several ideas aimed at overcoming the key obstacles to efficient POMDP planning: the "curse of dimensionality" and the "curse of history".  With computational tools getting better, POMDPs have been applied to a range of applications in robotics and beyond.  I will give tow examples on aircraft collision avoidance and on autonomous vehicle navigation.  Finally, I will connect planning and learning and discuss the application of POMDP planning to the learning question of dealing with model uncertainty.


Kris David Hsu is an Associate Professor of computer science at the National University of Singapore, a member of NUS Graduate School for Integrative Science & Engineering (NGS), and Assistant Dean of Graduate Students in the School of Computing.  His research spans robotics, AI, computational biology and geometric computation.

He received B.Sc in computer science and mathematics from the University of British Columbia, Canada and PhD in computer science from Stanford University, USA.  After leaving Stanford, he worked at Compaq Computer Corp's Cambridge Research Laboratory and the University of North Carolina at Chapel Hill.  At the National University of Singapore, he held teh Sung Kah Kay Assistant Professorship and was a Fellow of the Singapore-MIT Alliance.  He is currently serving on the editorial boards of IEEE Transactions on Robotics and Journal of Artificial Intelligence Research

Seminar Type: 

ITEE Research Seminar