Reproduce “RAGrasp: Geometry-Semantic Template Retrieval and Grasp Transfer” 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
Shenzhe Zhu, Chengxiao He, Jan Harder
RAGrasp adapts planar robot grasps to a deployment by retrieving locally annotated RGB-D examples rather than training a task-specific grasp network. It uses object masks, contour matching, DINOv2 features, and confidence-gated transfer with geometric fallback. In a 20-object unseen-object comparison, it succeeded on 19/20 (95%), versus NDF’s 17/20 (85%); performance depends on relevant template coverage and reliable masks and calibration.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.