Reproduce “SPAR: Search-trained Push Action Ranker for Efficient Navigation Among Movable Obstacles” 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
Dhruv Metha Ramesh, Tri-An Nguyen, Junshan Huang, Jingjin Yu, Kostas E. Bekris, Abdeslam Boularias
SPAR learns to prioritize robot pushes that open access through blocked workspaces. It uses simulated search outcomes to rank immediate openings above setup pushes that enable a later opening. On 300 simulated two-push keyhole problems, successful runs had median costs of five simulator calls for SPAR Search versus 35 for Random Search. These are success-conditioned medians, not success rates.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.