Reproduce “Geometry-Preserving Human-to-Robot Upper-Body Motion Retargeting from Monocular Video” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Xiaoyu Yang, Sen Han, Da Li, Nan Wu
The method reconstructs a person’s arms and hands as one connected articulated state from fixed-view RGB video, combining frame-wise, video-level, and hand estimates. It retargets upper-arm, forearm, wrist, elbow, and palm relationships to a robot’s kinematics rather than copying joint angles. Across 16 videos, hand reprojection error fell from 22.36 to 7.21 pixels, but this does not establish task success.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.