UofT Robotics: Gaurav Sukhatme (USC) on Big Data and Small Models: Lessons for Robotics скачать в хорошем качестве

UofT Robotics: Gaurav Sukhatme (USC) on Big Data and Small Models: Lessons for Robotics

Description: We have recently demonstrated the possibility of learning drone swarm controllers that are zero-shot transferable to real quadrotors via large-scale, multi-agent, end-to-end reinforcement learning. We train policies parameterized by neural networks that can control individual drones in a swarm in a fully decentralized manner. Our policies, trained in simulated environments with realistic quadrotor physics, demonstrate advanced flocking behaviors, perform aggressive maneuvers in tight formations while avoiding collisions with each other, break and re-establish formations to avoid collisions with moving obstacles, and efficiently coordinate in pursuit-evasion tasks. The model learned in simulation transfers to highly resource-constrained physical quadrotors performing station-keeping and goal-swapping behaviors. Motivated by these results and the observation that neural control of memory-constrained, agile robots requires small yet highly performant models, we have begun a project that leverages graph hypernetworks to learn hyperpolicies trained with off-policy reinforcement learning. Early results from this work on locomotion and manipulation tasks suggest that one can obtain performant networks that are two orders of magnitude smaller than those commonly used yet encode policies comparable to those encoded by much larger networks trained on the same task. The talk will conclude with some thoughts on the generality of such approaches for devices with modest computational capabilities. Bio: Gaurav S. Sukhatme holds the Fletcher Jones Foundation Endowed Chair in Computer Science at the University of Southern California (USC). He is a Professor of Computer Science and Electrical & Computer Engineering and serves as the Executive Vice Dean at the USC Viterbi School of Engineering. He is an Amazon Scholar. He received his undergraduate education at IIT Bombay in Computer Science and Engineering and M.S. and Ph.D. degrees in Computer Science from USC. He is the co-director of the USC Robotics Research Laboratory and the director of the USC Robotic Embedded Systems Laboratory, which he founded in 2000. His research interests are in networked robots, learning robots, and field robotics. He has published extensively in these and related areas. Sukhatme has served as PI on numerous NSF, DARPA, and NASA grants. He was a Co-PI on the Center for Embedded Networked Sensing (CENS), an NSF Science and Technology Center. He is a fellow of the AAAI and the IEEE and a recipient of the NSF CAREER and the Okawa Foundation research awards. He is one of the founders of the Robotics: Science and Systems conference and the Editor-in-Chief of Autonomous Robots.