Lockheed Martin Robotics Seminar: A dynamical systems framework for autonomous task management

Friday, October 30, 2020
2:00 p.m.
Online Zoom seminar. Registration is required.

Lockheed Martin Robotics Seminar

Registeration link:

https://umd.zoom.us/webinar/register/WN_GNphex58QHCcC3glspsZ6A

A dynamical systems framework for flexible, sensor-based autonomous task management

Paul Reverdy
Assistant Professor
Aerospace and Mechanical Engineering Department 
University of Arizona 

Abstract

Mobile robots are becoming increasingly ubiquitous due to technological advances in sensing, actuation, and computation. However, significant challenges remain in developing efficient ways to express tasks, and control strategies that ensure robots perform these tasks in unpredictable real-world environments. An approach that has proved useful in practice is to encode tasks as attractors of dynamical systems, e.g., design vector fields that steer a vehicle to a desired location. To encode richer behaviors, one can develop methods to switch among low-level point-attractor controllers; this is often done by developing an automaton that discretely switches between available controllers, yielding a hybrid dynamical system. In this talk, we describe an alternative approach based on a continuous dynamical systems mechanism for making decisions, i.e., choosing controllers. The system associates a value with each possible controller and then weights the control vector fields according to their values, with the highest value controller getting the highest weight. We show how the system encodes desirable behaviors, such as recurrent patrol and sensor-based task prioritization. We conclude the talk by 1) describing how these capabilities can be extended and composed within the framework, and 2) describing how the dynamical system can interface with human supervisors. 

Biography

Paul Reverdy received the B.S. degree in Engineering Physics and the B.A. degree in Applied Mathematics from the University of California, Berkeley in 2007, and the M.A. and Ph.D. degrees in Mechanical and Aerospace Engineering from Princeton University in 2011 and 2014, respectively. He is an Assistant Professor in Aerospace and Mechanical Engineering at the University of Arizona. Since July 2020, he has been at Amazon, working on navigation for the Scout delivery robot.

From 2007 to 2009, he worked as a research assistant at the Federal Reserve Board of Governors, Washington, DC. From 2014 to 2017, he was a postdoctoral fellow in Electrical and Systems Engineering at the University of Pennsylvania, where he was affiliated with the GRASP laboratory. His research interests lie at the intersection of human and machine decision making and control, with applications in robotics, machine learning, and engineering design optimization. Dr. Reverdy's awards include a National Defense Science and Engineering Graduate (NDSEG) Fellowship for graduate study and the best student paper award from the 2014 European Control Conference.

Host

Mumu Xu

Contact: appicard@umd.edu

remind we with google calendar

 

November 2020

SU MO TU WE TH FR SA
1 2 3 4 5 6 7
8 9 10 11 12 13 14
15 16 17 18 19 20 21
22 23 24 25 26 27 28
29 30 1 2 3 4 5
Submit an Event