AnyPos: Automated Task-Agnostic Actions for Bimanual Manipulation

ICLR 2026 Conference Submission19525 Authors

Keywords: Robot Learning, Computer Vision, Reinforcement Learning, Embodied AI

Abstract:

Learning generalizable manipulation policies hinges on data, yet robot manipulation data is scarce and often entangled with specific embodiments, making both cross-task and cross-platform transfer difficult. We tackle this challenge with \textbf{task-agnostic embodiment modeling}, which learns embodiment dynamics directly from \emph{task-agnostic action} data and decouples them from high-level policy learning. This data-driven perspective bypasses the limitations of traditional dynamics-based modeling and enables scalable reuse of action data across different tasks.
Building on this principle, we introduce \textbf{AnyPos}, a unified pipeline that integrates large-scale automated exploration with robust inverse dynamics learning. AnyPos generates diverse yet safe trajectories at scale, then learns embodiment representations by \textit{decoupling arm and end-effector motions} and employing a \textit{direction-aware decoder} to stabilize predictions under distribution shift, which can be seamlessly coupled with diverse high-level policy models.
In comparison to the standard baseline, AnyPos achieves a 51\% improvement in test accuracy. On manipulation tasks such as operating a microwave, toasting bread, folding clothes, watering plants, and scrubbing plates, AnyPos raises success rates by {30--40\%} over strong baselines. These results highlight data-driven embodiment modeling as a practical route to overcoming data scarcity and achieving generalization across tasks and platforms in visuomotor control.

Primary Area: applications to robotics, autonomy, planning

Submission Number: 19525