One-Step Flow Policy: Self-Distillation for Fast Visuomotor Policies

TL;DR

The paper introduces the One-Step Flow Policy (OFP), a self-distillation method that generates high-fidelity robot actions in a single step, significantly reducing inference latency and improving control in manipulation tasks.

Contribution

OFP is a novel self-distillation framework that enables fast, one-step action generation without a pre-trained teacher, combining self-consistency and regularization for high-quality control.

Findings

Outperforms 100-step diffusion policies in 56 manipulation tasks

Achieves over 100x faster action generation

Surpasses the original 10-step policy in RoboTwin 2.0

Abstract

Generative flow and diffusion models provide the continuous, multimodal action distributions needed for high-precision robotic policies. However, their reliance on iterative sampling introduces severe inference latency, degrading control frequency and harming performance in time-sensitive manipulation. To address this problem, we propose the One-Step Flow Policy (OFP), a from-scratch self-distillation framework for high-fidelity, single-step action generation without a pre-trained teacher. OFP unifies a self-consistency loss to enforce coherent transport across time intervals, and a self-guided regularization to sharpen predictions toward high-density expert modes. In addition, a warm-start mechanism leverages temporal action correlations to minimize the generative transport distance. Evaluations across 56 diverse simulated manipulation tasks demonstrate that a one-step OFP achieves state-of-the-art results, outperforming 100-step diffusion and flow policies while accelerating action generation by over $100\times$. We further integrate OFP into the $\pi_{0.5}$ model on RoboTwin 2.0, where one-step OFP surpasses the original 10-step policy. These results establish OFP as a practical, scalable solution for highly accurate and low-latency robot control.