Reproduce “Cross-Controller Transfer via In-Context Behavior Adaptation” 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
Jonathan Yang, Chelsea Finn, Dorsa Sadigh
The authors present a robot policy using examples of commands and resulting motion under a deployment controller, sometimes from another task. This context supplies evidence about controller-dependent expert behavior beyond estimated controller dynamics. On a five-task held-out task–controller evaluation, zero-shot average success was 50.0%, versus 27.0% for system identification; evidence comes from one lab setup and VLA architecture.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.