FLASH: Efficient Visuomotor Policy via Sparse Sampling

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bai, Jiaqi, Jia, Jindou, Hu, Yuxuan, Li, Gen, Chen, Xiangyu, An, Tuo, Zuo, Kuangji, Yang, Jianfei
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866917498894417920
author Bai, Jiaqi
Jia, Jindou
Hu, Yuxuan
Li, Gen
Chen, Xiangyu
An, Tuo
Zuo, Kuangji
Yang, Jianfei
author_facet Bai, Jiaqi
Jia, Jindou
Hu, Yuxuan
Li, Gen
Chen, Xiangyu
An, Tuo
Zuo, Kuangji
Yang, Jianfei
contents Generative models such as diffusion and flow matching have become dominant paradigms for visuomotor policy learning, yet their reliance on iterative denoising incurs high inference latency incompatible with real-time robotic control. We present Fast Legendre-polynomial Action policy via Sparse History-anchored flow (FLASH Policy), which replaces discrete action-chunk generation with continuous Legendre polynomial trajectory representation. Specifically, by fitting expert demonstrations under sparse temporal sampling, FLASH enables a single inference to cover a significantly extended action horizon. To further accelerate generation, FLASH initiates the flow matching process from history polynomial coefficients rather than uninformative Gaussian noise, shortening the transport distance and enabling accurate single-step inference. Moreover, analytic polynomial differentiation directly provides desired velocity feed-forward signals to the torque controller without numerical approximation. Extensive experiments on five simulated and two real-world manipulation tasks demonstrate that FLASH achieves state-of-the-art success rates ($\ge 92\%$ across all tasks), a per-episode inference time of $31.40\,ms$ (up to $175\times$ faster than diffusion policies and $18\times$ faster than prior flow matching policies), up to $4\times$ faster training convergence than ACT, and $5\times$ to $7\times$ reduction in controller tracking error compared to discrete-action baselines.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15492
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle FLASH: Efficient Visuomotor Policy via Sparse Sampling
Bai, Jiaqi
Jia, Jindou
Hu, Yuxuan
Li, Gen
Chen, Xiangyu
An, Tuo
Zuo, Kuangji
Yang, Jianfei
Robotics
Computer Vision and Pattern Recognition
Generative models such as diffusion and flow matching have become dominant paradigms for visuomotor policy learning, yet their reliance on iterative denoising incurs high inference latency incompatible with real-time robotic control. We present Fast Legendre-polynomial Action policy via Sparse History-anchored flow (FLASH Policy), which replaces discrete action-chunk generation with continuous Legendre polynomial trajectory representation. Specifically, by fitting expert demonstrations under sparse temporal sampling, FLASH enables a single inference to cover a significantly extended action horizon. To further accelerate generation, FLASH initiates the flow matching process from history polynomial coefficients rather than uninformative Gaussian noise, shortening the transport distance and enabling accurate single-step inference. Moreover, analytic polynomial differentiation directly provides desired velocity feed-forward signals to the torque controller without numerical approximation. Extensive experiments on five simulated and two real-world manipulation tasks demonstrate that FLASH achieves state-of-the-art success rates ($\ge 92\%$ across all tasks), a per-episode inference time of $31.40\,ms$ (up to $175\times$ faster than diffusion policies and $18\times$ faster than prior flow matching policies), up to $4\times$ faster training convergence than ACT, and $5\times$ to $7\times$ reduction in controller tracking error compared to discrete-action baselines.
title FLASH: Efficient Visuomotor Policy via Sparse Sampling
topic Robotics
Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2605.15492