Robot action planning made smoother with new movement tricks
Smooth Action Chunking Via Movement-Primitive Representations for Imitation Learning
Machine LearningRobotics
Summary
Robots that learn by watching can sometimes move jerkily because their action plans are made in chunks that don’t flow smoothly from one to the next. The authors invented a way to represent these action chunks using smooth building blocks that naturally make the robot’s motions gentle and connected. This method works with existing robot policies without retraining, cutting down the jerks a lot while keeping tasks just as successful. It’s been tested on both simulations and a real two-armed robot with good results.
What this means in practice
- •For robotics engineers: Enable robots to produce smoother, less jerky movements by integrating smooth trajectory primitives into their action planning systems.
- •For manufacturing automation teams: Reduce wear and vibration in robotic arms used in factories by applying the method to improve motion smoothness without retraining control policies.
Authors
Shota Nomura, Takayuki Osa, Yasuhiro Kato, Kazuki Ota, Naoki Shitanda, Tatsuya Harada
Abstract
Imitation learning has become a widely adopted approach for vision-based robot policies, owing to its conceptual simplicity, scalability, and ability to leverage large-scale datasets. As these policies have grown more capable, action chunking has become their standard interface. At each decision step the policy outputs several future actions at once. Backbones from diffusion policies to large vision-language-action models all share it. Whatever backbone produces the chunk, its temporal structure is learned only implicitly. The commanded trajectory is therefore jerky both within a chunk and between successive chunks. On real hardware this appears as visible vibration and as repeated saturation of the follower's velocity limit. We address this issue by incorporating a movement-primitive-based trajectory representation into action chunking. Specifically, a trajectory segment is represented as a linear combination of smooth basis functions, yielding smooth trajectories by construction while leaving the policy architecture, training objective, and inference procedure unchanged. The same representation also yields a closed-form, inference-time correction for the second source of jerk. We condition the predicted coefficients so that the new chunk passes through a via-point at the previous chunk's last commanded action. This correction needs no retraining and, unlike asynchronous Real-Time Chunking, applies to policies queried in one shot. We instantiate it on a one-shot (ACT) and an iterative (DiT) backbone and evaluate in simulation and on a real bimanual ALOHA platform. The representation lowers within-chunk jerk by an order of magnitude on the one-shot backbone and by 2.4 times on the iterative one, with task success statistically non-inferior. The conditioning then removes 73-84% of the boundary discontinuity, at negligible per-step cost.