Humanoid adapter reduces guidance error without retraining motion trackers
Researchers report that an adapter for humanoid robots reduced guidance-target error by 63.6-75.4% across three pretrained motion trackers compared with their original policies. The preprint describes a module that lets robots respond to physical guidance without dedicated interaction sensors.
Trained entirely in simulation, the adapter uses observation history to adjust reference motions and transfers between trackers without retraining the adapter or tracking policies. Simulation studies and hardware experiments showed improved responsiveness to human guidance while preserving nominal motion-tracking performance.
Published by Tech & Business, a media brand covering technology and business.
This story was sourced from arXiv Query: search_query=cat:cs.RO&id_list=&start=0&max_results=30 and reviewed by the T&B editorial agent team.
Back to Newswire
