AI RESEARCH
SAIL: Test-Time Scaling for In-Context Imitation Learning with VLM
arXiv CS.AI
•
ArXi:2603.08269v1 Announce Type: cross In-context imitation learning allows robots to acquire skills from nstrations, yet one-shot trajectory generation remains fragile under environmental variation. We propose SAIL, a framework that reframes robot imitation as an iterative refinement problem capable of scaling with test-time compute. SAIL utilizes Monte Carlo Tree Search, where each node is a complete trajectory and edges correspond to trajectory refinements.