Reproduce “AcTExplore: Active Tactile Exploration of Unknown Objects” 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
Amir-Hossein Shahidzadeh, Seong Jong Yoo, Pavan Mantripragada, Chahat Deep Singh, Cornelia Fermüller, Yiannis Aloimonos
AcTExplore trains a robot finger to map reachable surfaces of unknown rigid objects through touch. Its policy uses recent tactile depth readings and rewards contact area and less-visited contact locations. It is trained on cube and sphere primitives, then tested on unseen YCB objects. The authors report 95.97% average coverage under a 5 mm surface-distance metric; scissors reaches 74.62%.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.