Falcon: Fast Visuomotor Policies via Partial Denoising

Haojun Chen, Minghao Liu, Chengdong Ma, Xiaojian Ma, Zailin Ma, Huimin Wu, Yuanpei Chen, Yifan Zhong, Mingzhi Wang, Qing Li, Yaodong Yang

Abstract: Diffusion policies are widely adopted in complex visuomotor tasks for their ability to capture multimodal action distributions. However, the multiple sampling steps required for action generation significantly harm real-time inference efficiency, which limits their applicability in real-time decision-making scenarios. Existing acceleration techniques either require retraining or degrade performance under low sampling steps. Here we propose Falcon, which mitigates this speed-performance trade-off and achieves further acceleration. The core insight is that visuomotor tasks exhibit sequential dependencies between actions. Falcon leverages this by reusing partially denoised actions from historical information rather than sampling from Gaussian noise at each step. By integrating current observations, Falcon reduces sampling steps while preserving performance. Importantly, Falcon is a training-free algorithm that can be applied as a plug-in to further improve decision efficiency on top of existing acceleration techniques. We validated Falcon in 48 simulated environments and 2 real-world robot experiments. demonstrating a 2-7x speedup with negligible performance degradation, offering a promising direction for efficient visuomotor policy design.

Lay Summary: Robots using advanced "diffusion policies" can perform complex tasks flexibly, but they're too slow for real-world use because they need many computational steps to make each decision. We developed Falcon, a new method that makes robots act faster by cleverly reusing information from their previous actions. Unlike existing solutions that sacrifice performance for speed, Falcon maintains the robot's ability to perform tasks successfully while making decisions 2-7 times faster. Our method works as a simple plug-in to existing robot systems without requiring additional training. We proved Falcon's effectiveness across 48 simulated tasks and 2 real-world robot experiments, showing it can help robots respond more quickly in practical applications.

Link To Code: https://github.com/chjchjchjchjchj/Falcon

Primary Area: Applications->Robotics

Keywords: Diffusion Models, Acceleration, Sampling, Robotics, Motion Planning

Submission Number: 506