Siheng Zhao

A new method automates dense reward functions using large language models.

A system called TieBot helps robots learn to knot ties using simulations.

Transforming robot training through human-like movement captured on video.