Reproduce “Cross-Embodiment Affordance Transfer for Robotic Manipulation” 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

Sawera Yaseen, Serge Thill, Rekha Raja

The authors condition point-cloud affordance predictions on a robot’s geometry and use simulated execution trials to adapt human-annotated grasp and wrap-grasp labels. In simulation, the model reaches 53.64% macro mean intersection-over-union, versus 52.46% without robot information. This modest gain is measured on simulation-derived labels for two grippers sharing an arm, not real-world task success.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.