GitHub

Closed-Loop Visuomotor Control with Generative Expectation for Robotic Manipulation

🍀 CLOVER employs a text-conditioned video diffusion model for generating visual plans as reference inputs, then these sub-goals guide the feedback-driven policy to ...