Ambient Diffusion Policy

Ambient Diffusion Policy:
Imitation Learning from Suboptimal Data in Robotics

Adam Wei, Nicholas Pfaff, Thomas Cohn, Arif Kerem Dayı, Constantinos Daskalakis, Giannis Daras, Russ Tedrake

arXiv Code coming soon YouTube coming soon
🚧 Website under construction

This project page is still being built. The full method, results, and paper links are coming soon. Please check back shortly.

Example Hardware Rollouts

Example real-world rollouts from the hardware experiments.

Table cleaning: Ambient

Table cleaning: co-training

Tower building: Ambient

Tower building: co-training