Abstract: Learning from demonstration is a powerful method for teaching robots new skills, and having more demonstration data often improves policy learning. However, the high cost of collecting demonstration data is a significant bottleneck. Videos, as a rich data source, contain knowledge of behaviors, physics, and semantics, but extracting control-specific information from them is challenging due to the lack of action labels. In this work, we introduce a novel framework, \textbf{A}ny-point \textbf{T}rajectory \textbf{M}odeling (ATM), that utilizes video demonstrations by pre-training a trajectory model to predict future trajectories of arbitrary points within a video frame. Once trained, these trajectories provide detailed control guidance, enabling the learning of robust visuomotor policies with minimal action-labeled data. Across the \textbf{130} language-conditioned tasks we evaluated in both simulation and the real world, ATM outperforms strong video pre-training baselines by 80$\%$ on average. Furthermore, we show effective transfer learning of manipulation skills from human videos.
