LMSS @ Cornell Tech: Danfei Xu (Georgia Tech)

Learning Machines Seminar Series What: LMSS: Danfei Xu (Georgia Tech) When: Monday, October 5, 12:00 p.m. - 1:00 p.m. Where: Bloomberg 091 The series is organized by Associate Professor Yoav Artzi and sponsored by Bloomberg. Pizza will be served at 11:45 a.m. "Robots That Inherit Human Experience" The world we want robots to enter is a human world, shaped by our bodies, goals, and accumulated physical knowledge. Yet robot learning still relies largely on scarce demonstrations collected through teleoperation. Can robots instead inherit physical intelligence from the rich experience people generate every day? The challenge is not merely to copy human motion, but to extract knowledge that a different body can use. I will present three lines of work toward this goal: EgoScale and CHORD, which explore scalable learning for dexterous manipulation; EgoWAM, which learns transferable physical dynamics through world action modeling; and WARP, which transfers whole-body human motion and visuomotor skills to mobile manipulators and humanoids. Together, these projects ask what knowledge about physical intelligence is shared across embodiments, and what representations and transformations make that knowledge usable by a machine. BIO Danfei Xu is an Assistant Professor in the School of Interactive Computing at Georgia Tech, where he directs the Robot Learning and Reasoning Lab, and a Research Scientist at NVIDIA. His research lies at the intersection of robotics and machine learning, with a focus on learning generalizable robot skills from human experience. He received his Ph.D. from Stanford University and his B.S. from Columbia University. His honors include the NSF CAREER Award and the ICRA Best Conference Paper Award.