CSL Master Calendar

Back to Listing

DCL Lecture Series: Jonathan How "Planning and Learning for Multiagent Teams"

Event Type
Seminar/Symposium
Sponsor
Decision and Control Laboratory, Coordinated Science Laboratory
Location
CSL Auditorium (B02)
Date
Apr 30, 2014   3:00 - 4:00 pm  
Speaker
Prof. Jonathan How, Massachusetts Institute of Technology
Contact
Angie Ellis
E-Mail
amellis@illinois.edu
Phone
217/30-1910
Views
346
Originating Calendar
CSL Decision and Control Group

Title: Planning and Learning for Multiagent Teams

Abstract: This talk will present algorithmic solutions to some key challenges in autonomous learning and planning. In single agent domains, reliable robust planning depends on the availability of models that accurately represent the environment and its evolution.

One approach for efficiently learning probabilistic models in dynamic environments utilizes Bayesian nonparametric models that provide the flexibility to learn both the structure and the parameters of a model, which are often very difficult to determine a priori. For example, in motion planning domains, Gaussian processes (GPs) are used to represent the trajectory velocity fields of obstacles (static & dynamic) in the environment, a Dirichlet process GP mixture (DP-GP) is used to learn the number of motion models and their velocity fields, and the dependent Dirichlet process GP mixture (DDP-GP) augments these models by capturing the obstacles' temporal evolution. In practice, these techniques have been used to learn models of motion/intent behaviors of automobiles and pedestrians to improve the performance of autonomous vehicles.

When moving to multiagent planning and learning, other challenges are introduced when limited communication, due to hardware constraints or features of the operation environment, do not allow for centralized learning and planning. Learning in these environments forces individual agents to construct and share compact representations of data rich observations so that manageable communication levels can be achieved while allowing for model consistency across agents.

Likewise for planning, agents need to be explicitly aware of, and account for, the effect of limits in inter-agent bandwidth on planning. This talk will present algorithms for both aspects of decentralized multiagent operations.

Biography:

Dr. Jonathan P. How is the Richard C. Maclaurin Professor of Aeronautics and Astronautics at the Massachusetts Institute of Technology.  He received a B.A.Sc. from the University of Toronto in 1987, S.M. & Ph.D. from MIT Aeronautics and Astronautics in 1990 and 1993, respectively, and then was a postdoctoral associate for two years at MIT. Prior to returning to MIT in 2000, he was an Assistant Professor in the Department of Aeronautics and Astronautics at Stanford University. He is the Deputy Editor-in-Chief of the IEEE Control Systems Magazine and an Associate Editor for the AIAA Journal of Aerospace Information Systems. Professor How was the recipient of the 2002 Institute of Navigation Burka Award, a Boeing Special Invention award in 2008, the 2011 IFAC Automatica award for best applications paper, Recipient of the AIAA Best Paper Award from the 2011 and 2012 Guidance Navigation and Control Conferences, and he is an Associate Fellow of AIAA and a senior member of IEEE.

link for robots only