Search papers, labs, and topics across Lattice.
MIT CSAIL2077AI Open Source FoundationAffiliation:NorthwesternTexas A&MUW-MadisonSep 29, 2026arXiv:2609.37359ENCORE enables few-shot agentic discovery of manipulation strategies by distilling demonstrations into multi-view keyframes and events that a coding agent uses to write and iteratively refine a policy program. This approach allows agents to learn complex robotic tasks from limited examples, achieving high success rates on perturbed tasks and outperforming prior agentic systems. The method is validated on simulated benchmarks and a real bimanual robot, demonstrating its effectiveness for tasks with underspecified instructions.
Robotic agents can now learn complex manipulation strategies from just a few demonstrations by treating them as evidence to be analyzed, not just training data.
Coding agents can now write, run, and debug programs with little human help. Robot tasks, however, are usually specified by a sentence that leaves out how to grasp, in what order to make contact, and what the result should look like, and an agent given only the sentence must find these details by trial and error. We introduce ENCORE, which gives the agent a few demonstrations as evidence to read rather than as training data. A deterministic builder distills each demonstration into a pack of multi-view keyframes, gripper events, frame strips, and the full trajectory. A coding agent studies the pack, writes a policy program against a fixed perception and action API, refines it iteratively over a few development rollouts, and freezes it before a sealed evaluation that never reveals the success signal. On LIBERO-PRO, the agent's first program already succeeds in half of the perturbed tasks with demonstrations and in one task without them, and the frozen programs outperform the strongest prior agentic system run with the same language model (96.3% against 89.3%). On RoboDojo tasks whose instructions leave the goal unstated, no program succeeds without demonstrations. ENCORE also runs on a real bimanual robot, learning cube handover and cup inversion from five demonstrations each.
