Reproduce “On Agentic Manipulation Policies” 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
Chen Chen, Omid Bagherieh, Naveen Nikamanth Appiah Balaji, Jesse T. Gonzalez, Negin Heravi, Fahad Islam, Xueting Li, Jean-Baptiste Regli and 5 more
AMP1 lets a general-purpose model control robot manipulation through code that observes the scene, commands motion, checks outcomes, and revises actions. It stays involved as tasks unfold, rather than relying on skill menus or programs without feedback. On RoboLab-120, AMP1 with Gemini 4 Argon succeeds in 85.6% of episodes; physical trials cover one single-arm Franka setup, and the system is slow.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.