Reproduce “Active Perception for Embodied Disambiguation” 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
Yiwei Liu, Luwei Yang
The authors present a robot policy in which a vision-language model decides whether to obtain a new RGB-D view, ask a clarification question, or select a target. New viewpoints can reveal hidden or out-of-view evidence and improve later questions. Across 9 matched real-robot scenarios, the method was marked correct in all 9, but this small study does not establish generalization.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.