InterMimicGen: Scaling Humanoid Loco-Manipulation through Self-Evolving Motion Imitation
InterMimicGen grows humanoid loco-manipulation data by retargeting human demos and self-evolving a physics tracker.
InterMimicGen retargets heterogeneous motion-captured human-object interactions into humanoid references that preserve whole-body coordination and hand-object contact. A physics-based generalist tracker executes those references in simulation on a dexterous humanoid. A data flywheel then applies small task-preserving edits, fine-tunes the tracker, and retains only variants whose simulated execution succeeds. The authors report growing executable coverage and transfer to real robots.
- Retargets human-object captures into coordinated humanoid references
- Trains one generalist physics tracker for dexterous loco-manipulation
- Flywheel keeps only task-successful motion variants each round
- Reports broader coverage and transfer to real robots
From the original article
1 min read at huggingface.co
Captured human-object interactions provide rich supervision for humanoid loco-manipulation, but they are sparse, heterogeneous, and not directly executable by robots. We introduce InterMimicGen, a self-evolving motion-imitation framework in which robot motion data and a tracking policy improve each other.
By Yucheng Zhang, Sirui Xu, Jinhong Li · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.