LocomotionHardware: MaxRoadmap scaffold
Humanoid locomotion fine-tune
Pull a foundation walking policy, fine-tune in sim, deploy to G1.
Lab runtime / LocomotionOK / VIS-01
Humanoid locomotion fine-tune runs against the same observable edge stack used for testing and deployment.
This entry is a roadmap scaffold, not a finished demo. Its description records intended scope; the linked repository is the source of truth for current code.
About this demo
Pull a foundation locomotion policy from the Model Hub, fine-tune for your terrain in cloud simulation with PPO + domain randomization, and deploy to a humanoid in under an hour. Cross-embodiment transfer between G1/H1.
Highlights
- →Cross-embodiment transfer
- →Domain-randomized PPO post-training
- →ONNX export to Jetson
- →Foot-contact telemetry overlay
Supported robots
Unitree G1Unitree H1Stanley H1