Reproduce “From a Single RGB Image to Robot Execution: Metric Scene Alignment and Contact-Guided Retargeting” 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
Zhenxuan Ding, Yu Huang, Xinmiao Wang, Chenghao Sun, Ziyang Dong, Wenhao Cui
The authors present a reference-generation pipeline that turns one RGB image into humanoid interaction references for support-dominant tasks. SAM 3D reconstructs the scene; JMIR aligns a generated human motion to it, and GMR transfers the motion to a robot before CGRR corrects support errors. In simulation, the pipeline reaches 89% success over 100 trials, but only in tested settings.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.