Runpei Dong

Robots can now achieve precise physical interactions with objects without relying on task-specific rewards, thanks to a novel approach that decouples contact from keypoint tracking.

Humanoids can now perform complex loco-manipulation tasks from egocentric vision and sparse goals, thanks to a unified controller trained without relying on predefined motion references at test time.

Humanoid robots can now manipulate everyday objects in the wild with unprecedented reliability, thanks to a novel system that combines the generalization of large vision models with a highly accurate, simulation-trained end-effector tracking policy.