Robot motion planning improves with compact joint-space sketches
Proprioceptive Sketches as Long-Horizon Intent for Generative Action Policies
Robotics
Summary
Robots often plan short movements without a clear long-term goal, which can make their actions less efficient. The authors present a new method that creates a simple, timing-free outline of the robot’s future movements, capturing the essential shape of the path it needs to follow. This outline helps guide detailed actions, making robot tasks more successful. Their approach showed better results in both simulations and real-world robot tasks than methods that predict actions alone.
What this means in practice
- •For robotics engineers: Design robot controllers that predict compact joint movement sketches to improve task success rates in long and complex operations.
- •For automation system integrators: Integrate compact long-horizon motion intent into robot workflows for more reliable operation in manufacturing or assembly tasks.
Authors
Fangyuan Wang, Songhao Huang, Haoxiang Sun, Shipeng Lyu, Chengyang He, Anqing Duan, Peng Zhou, David Navarro-Alarcon
Abstract
Generative robot policies predict short action chunks but lack explicit long-horizon intent. Recent methods expose longer-horizon structure through language plans, subgoal images, or video forecasts, which are costly to generate and still need to be translated into robot motion. Predicting future robot motions avoids this translation, but a dense, time-indexed trajectory requires numerous parameters to cover the full remaining task, and over a short horizon it largely repeats the action chunk and adds little guidance for action generation. We propose Proprioceptive Action Models (PAM), which jointly generate a compact, timing-free sketch of the robot's remaining joint-space path and a dense executable action chunk within a single transformer denoiser. The sketch parameterizes the path by arc length rather than time, capturing geometric intent invariant to execution timing. Block-causal attention and a staggered denoising schedule maintain directed sketch-to-action dependence, ensuring the action tokens condition on a progressively cleaner sketch throughout sampling. In simulation, PAM improves over its action-only counterparts on Push-T and LIBERO-Long; on four real-world bimanual tasks, it raises success from 47.5% to 75.0%. Project page: https://nicehiro.github.io/pam_dp/