Robot policies

Every policy here was trained in RL Studio from a description or a reference video, and exported as ONNX with its observation normalizer baked in — download one and run it on the robot. Hover a card to watch the policy.