Motion Tracks: A Unified Representation for Human-Robot Transfer in Few-Shot Imitation Learning
This addresses the problem of scalable robot skill acquisition for robotics researchers and practitioners by enabling few-shot imitation learning from human videos, though it is incremental in leveraging cross-embodiment action spaces.
The paper tackles the challenge of training robots via imitation learning without extensive robot demonstrations by proposing a unified action representation called motion tracks, which are short-horizon 2D trajectories on images, and achieves an average success rate of 86.5% across 4 real-world tasks, outperforming baselines by 40%.
Teaching robots to autonomously complete everyday tasks remains a challenge. Imitation Learning (IL) is a powerful approach that imbues robots with skills via demonstrations, but is limited by the labor-intensive process of collecting teleoperated robot data. Human videos offer a scalable alternative, but it remains difficult to directly train IL policies from them due to the lack of robot action labels. To address this, we propose to represent actions as short-horizon 2D trajectories on an image. These actions, or motion tracks, capture the predicted direction of motion for either human hands or robot end-effectors. We instantiate an IL policy called Motion Track Policy (MT-pi) which receives image observations and outputs motion tracks as actions. By leveraging this unified, cross-embodiment action space, MT-pi completes tasks with high success given just minutes of human video and limited additional robot demonstrations. At test time, we predict motion tracks from two camera views, recovering 6DoF trajectories via multi-view synthesis. MT-pi achieves an average success rate of 86.5% across 4 real-world tasks, outperforming state-of-the-art IL baselines which do not leverage human data or our action space by 40%, and generalizes to scenarios seen only in human videos. Code and videos are available on our website https://portal-cornell.github.io/motion_track_policy/.