Woodruff School of Mechanical Engineering

Faculty Candidate Seminar

Title:

Closing the Loop on Uncertainty Propagation and Autonomous Systems Control

Speaker:

Dr. Jonathan Rogers

Affiliation:

Texas A&M University

When:

Monday, January 14, 2013 at 11:00:00 AM

Where:

MRDC Building, Room 4211

Host:

Bill Singhose
bill.singhose@me.gatech.edu

Abstract

Today’s robots are still largely limited in the complexity of tasks they can perform and do not approach the reasoning ability of humans. These limitations are in part due to our inability to translate the fundamental decision-making processes used by humans to evaluate complex tradeoffs under uncertainty. As a result, current robotic systems often require undue operator oversight. This seminar will cover recent advances in stochastic control and automation that couple uncertainty propagation with optimal control. Recently derived mathematical techniques for nonlinear uncertainty propagation now allow robotic systems to predict uncertain future effects of their actions in real-time. Optimal control laws based on this predicted uncertainty then allow them to determine proper control inputs more effectively. The coupling between uncertainty propagation and optimal control allows autonomous systems to hedge their bets against uncertainty and protect themselves from unexpected scenarios that reduce performance or lead to mission failure. Examples are provided in the form of autonomous parafoil and guided projectile flight controllers that make use of real-time uncertainty propagation for obstacle avoidance in constrained landing environments. Results show that so-called “uncertainty-aware” control formulations far outperform deterministic control in the presence of realistic model error, stochastic forcing, and sensing uncertainty.


Biography

Dr. Jonathan Rogers is an Assistant Professor of Aerospace Engineering at Texas A&M University and director of the Helicopter and Unmanned Systems Laboratory. His research areas of specialization include stochastic control, nonlinear estimation, and alternative navigation technologies for low-cost autonomous systems. Dr. Rogers received his MS and PhD degrees in Aerospace Engineering from Georgia Tech in 2007 and 2009 respectively, and a BS in Physics from Georgetown University in 2006. His research has been funded by the US Army Research Laboratory, Army Research Office, and industry sponsors including Northrop Grumman, General Dynamics, SAIC, and BAE Systems. Dr. Rogers received the Army Research Office Young Investigator Award in 2012 for his work in state estimation of complex dynamical systems.

Notes

Refreshments will be served.