Learning From Out-Of-Distribution Data In Robotics

Abstract

In this talk, I present a general and principled framework for training generative models from low-quality or out-of-distribution data. I focus primarily on applications in robot imitation learning. This talk covers my most recent on-going research (as of Sept 2025).

Adam Wei

EECS Ph.D. Candidate

Robotics PhD student at MIT CSAIL. I am interested in robotics, imitation learning, generative modeling, and controls.