arXiv Robotics — research abstracts· Xingyu Zhang, Yingchao Li, Peizheng Guo, Jie Hu, Changwen Zheng, Wenwen Qiang·· 3 days agoEditorial score55
Cross-Embodiment Causal Transfer from Human Videos to Robot Manipulation
Cross-Embodiment Causal Transfer from Human Videos to Robot Manipulation
Summary
This research explores how to transfer knowledge from human manipulation videos to robot manipulation by analyzing subtask transfer through causal beams. The proposed method, Causal-Beam-Guided Subtask Transfer (CBST), learns manipulation-related representations from human videos and grounds them in task-level semantic variables using robot demonstrations. Experiments show improvements in task success over baseline world-action models.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
Source：arXiv Robotics — research abstracts · arxiv.org