RaC: Robot Learning for Long-Horizon Tasks by Scaling Recovery and Correction
This addresses inefficiencies in robot learning for complex, long-horizon tasks like bimanual control, offering a scalable method to improve robustness and efficiency, though it is incremental as it builds on existing imitation learning paradigms.
The paper tackles the problem of robot imitation learning plateauing on long-horizon tasks by introducing RaC, a training phase that fine-tunes policies on human intervention data for recovery and correction, resulting in outperforming prior state-of-the-art with 10x less data collection time and samples across real-world and simulated tasks.
Modern paradigms for robot imitation train expressive policy architectures on large amounts of human demonstration data. Yet performance on contact-rich, deformable-object, and long-horizon tasks plateau far below perfect execution, even with thousands of expert demonstrations. This is due to the inefficiency of existing ``expert'' data collection procedures based on human teleoperation. To address this issue, we introduce RaC, a new phase of training on human-in-the-loop rollouts after imitation learning pre-training. In RaC, we fine-tune a robotic policy on human intervention trajectories that illustrate recovery and correction behaviors. Specifically, during a policy rollout, human operators intervene when failure appears imminent, first rewinding the robot back to a familiar, in-distribution state and then providing a corrective segment that completes the current sub-task. Training on this data composition expands the robotic skill repertoire to include retry and adaptation behaviors, which we show are crucial for boosting both efficiency and robustness on long-horizon tasks. Across three real-world bimanual control tasks: shirt hanging, airtight container lid sealing, takeout box packing, and a simulated assembly task, RaC outperforms the prior state-of-the-art using 10$\times$ less data collection time and samples. We also show that RaC enables test-time scaling: the performance of the trained RaC policy scales linearly in the number of recovery maneuvers it exhibits. Videos of the learned policy are available at https://rac-scaling-robot.github.io/.