The School of Informatics and Computing (SoIC) Computer Science (CS) Colloquium Series
Speaker: Siddharth Srivastava, Staff Research Scientist
Where: Lindley Hall, Rm. 102
When: Friday March 3, 2017, 3:00 pm
Topic: Sequential Decision Making in the Real World: Representation, Computation, and Execution
Abstract: Intelligent assistive agents hold the key to making preventative precision healthcare a reality. Not only could such agents provide easily accessible assistants for the elderly and the ailing, but they could also assist by collecting valuable lifestyle information and making health-oriented suggestions without requiring households to be retrofit with ubiquitous sensors. In order to be truly helpful however, such agents would need to solve the problem of sequential decision making: determining what they should do in order to accomplish a designated objective. In this talk I will discuss some of the technical gaps that have been recognized as critical in solving such problems, along with our approaches towards resolving them.
I will show that combining the strengths of logical and probabilistic representations allows us to effectively express and solve large sequential decision making problems that could not be expressed using existing approaches. This synthesis draws upon fundamental ideas from decision theory, probabilistic inference and first-order modal logic. I will also present some of our work in closing the loop between planning and execution in robotics. We show that a principled integration of discrete planning with continuous planning allows present-day robots such as the PR2 to carry out assistive household tasks such as doing the laundry and setting a table for dinner.
Biography: Siddharth Srivastava is a Staff Scientist at the United Technologies Research Center in Berkeley. Prior to that, he was a postdoctoral researcher in the RUGS group (headed by Stuart Russell) at the University of California Berkeley. He received his PhD from the University of Massachusetts Amherst in 2010, working with Shlomo Zilberstein and Neil Immerman. His research interests include robotics and AI, with a focus on planning, reasoning and acting under uncertainty. His work on integrated task and motion planning for household robotics has received coverage from international news media. His dissertation work received a Best Paper award at the International Conference on Automated Planning and Scheduling (ICAPS) and an Outstanding Dissertation award from the Department of Computer Science at UMass Amherst.