Reproduce “AWAM: Active Trajectory-Conditioned View Imagination for Robust Robot Manipulation” 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

Jiawei Geng, Zhenyang Liu, Yuzheng Wu, Yanwei Fu

AWAM helps robots act when task-relevant objects or relationships are hidden by predicting task-conditioned virtual camera views of the initial scene. It combines geometry-conditioned visual predictions from those views with live camera observations in a fixed memory used by a closed-loop policy. On LIBERO-Occ, it averages 75.9% success versus 74.0% for VIM, but trails Fast-WAM on Long, 41.0% to 44.2%.

Summary by alphaXiv.

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