Reproduce “VeriGraph: Scene Graphs for Execution Verifiable Robot Planning” 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

Daniel Ekpo, Mara Levy, Saksham Suri, Chuong Huynh, Archana Swaminathan, Abhinav Shrivastava

VeriGraph represents tabletop scenes and goals as graphs of objects and spatial relations, then uses graph edits to propose, check, and revise manipulation plans. A vision-language model generates the graphs, while GPT-4 plans actions and validates preconditions iteratively. On reference-image block rearrangement, success was 0.86 versus 0.67 direct planning, but checks rely on image-inferred graphs and physical execution was open-loop.

Summary by alphaXiv.

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