TERRA: Terrain-Aware Reconstruction, Retargeting and Control for Musculoskeletal Locomotion
Researchers present TERRA, a pipeline that retargets motion data for muscle-driven locomotion on non-flat terrain.
Researchers present TERRA, a pipeline that reconstructs task-relevant terrain from kinematic motion alone and retargets it onto musculoskeletal bodies. It uses terrain priors, estimated contacts, negative free-space evidence, and anatomical, tendon, and contact constraints. A single muscle-actuated policy trained on 9.4 hours from five datasets improved terrain accuracy, cut interaction violations, and posted the highest completion rate on held-out terrain families.
- TERRA recovers support geometry from kinematics, terrain priors, contacts, and free-space evidence.
- Retargeting enforces anatomical, tendon-continuity, and contact constraints.
- One muscle-actuated policy was trained on 9.4 hours from five datasets.
- It improves terrain accuracy and reports the highest completion rate on supported terrain.
Full article161 words · extracted from huggingface.co · click to collapse
Recent advances in musculoskeletal modeling and reinforcement learning have enabled muscle-actuated agents to reproduce increasingly complex human motions. Yet these capabilities remain largely confined to flat ground, in part because motion datasets rarely include aligned terrain geometry and because retargeting terrain interactions to complex musculoskeletal bodies is challenging. We present TERRA, an end-to-end pipeline for terrain-aware retargeting and control of musculoskeletal locomotion. From kinematic trajectories alone, TERRA combines terrain priors, estimated contacts, and negative free-space evidence to recover task-relevant support geometry. TERRA further considers anatomical, tendon-continuity, and contact constraints during retargeting. Using the resulting motion-terrain pairs from five datasets, we successfully train a single muscle-actuated control policy on 9.4 hours of diverse locomotion. Across reconstruction, retargeting, and held-out tracking benchmarks, TERRA improves terrain accuracy, sharply reduces anatomical and interaction violations, and achieves the highest observed completion rate over supported terrain families. Overall, TERRA provides a practical route from scene-less motion data to muscle-actuated locomotion over diverse non-flat terrain. Project website: https://cnai.epfl.ch/terra/
Text extracted automatically; images, tables and formatting may be missing. Original: https://huggingface.co/papers/2609.38653