Speaker: Fangchen Liu UC Berkeley
Time: 2024-12-19 10:00-2024-12-19 11:00
Venue: 线上 (https://meeting.tencent.com/dm/Fw8lBQBQAtjp)

Abstract:

Specialist robots excel in many industrial tasks but remain limited to constrained environments and pre-defined objectives, making it hard for deployment in the unstructured physical world. Generalist robots, however, aim to handle free-form tasks across open-ended settings. In this talk, I will share our recent progress toward developing such generalist capabilities, focusing on (1) building models for flexible contextual task understanding, (2) enhancing the open-endedness and generalizability with pre-trained representations, and (3) collecting and leveraging diverse data sources effectively.

Short Bio:

Fangchen Liu is a fifth-year Ph.D. student at UC Berkeley, advised by Prof. Pieter Abbeel. Prior to that, she obtained an M.S. from UC San Diego working with Prof. Hao Su, and a B.S. from Peking University. Her current research focuses on developing algorithms and systems to improve the generalizability and dexterity of robotic manipulation. In the past, she also spent time at FAIR, NVIDIA Research, and Google Brain, where she worked on reinforcement learning algorithms and their integration with foundation models.