Note: This content is accessible to all versions of every browser. However, this browser does not seem to support current Web standards, preventing the display of our site's design details.


Coupling complex models and decision-making in robotics

Model learning is critical in many robotic systems - forming an abstract/compact representation of the world is necessary for prediction, decision-making, and keeping both computation and data complexity low. This learning can follow the conventional (decoupled) approach in which planning is based on a model learned offline, but when an appropriate model structure/set of observations to make is not known a priori, coupled model learning and decision-making can achieve better performance. This talk will explore some recent developments in advanced statistical modeling in the context of robotics, and in coupled modeling and decision-making for both single and multiagent systems. In the realm of single agent systems, the talk will cover work on information efficient map building using feature graphs, which is particularly applicable for robots with RGB-D sensors. The talk will also discuss multiagent modeling and decision-making, focusing on a novel streaming, distributed, asynchronous inference framework for Bayesian nonparametrics. Recent results on decentralized partially observable semi-Markov decision process (Dec-POSMDP), which uses macro-actions to facilitate solutions for large continuous-space decision-making problems will also be covered. The talk will showcase numerous result videos, including a multi-robot package delivery experiment conducted in our augmented-reality motion-capture facility.
Collaboration with Beipeng Mu, Miao Liu, Trevor Campbell, Shayegan Omidshafiei, Shih-Yuan Liu

Type of Seminar:
Control Seminar Series
Prof. Jonathan P. How
Massachusetts Institute of Technology
Oct 12, 2015   16:15

Contact Person:

Prof. Florian Dörfler
File Download:

Request a copy of this publication.
Biographical Sketch:
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 and his S.M. and Ph.D. in Aeronautics and Astronautics from MIT in 1990 and 1993, respectively. He then studied for two years at MIT as a postdoctoral associate for the Middeck Active Control Experiment (MACE) that flew onboard the Space Shuttle Endeavour in March 1995. Prior to joining MIT in 2000, he was an Assistant Professor in the Department of Aeronautics and Astronautics at Stanford University. He is the editor-in-chief of the IEEE Control Systems Magazine and an Associate Editor for the AIAA Journal of Aerospace Information Systems. Professor How received the 2002 Institute of Navigation Burka Award, the 2011 IFAC Automatica award for best applications paper, and AIAA conference best paper awards in 2011, 2012 and 2013. His team was recently awarded first place in the 2015 IEEE CSS Video Clip Contest.