One-Step Flow Policy: Self-Distillation for Fast Visuomotor Policies
This addresses the need for fast, high-precision robot control in time-sensitive manipulation, representing a strong specific gain rather than a foundational advancement.
The paper tackled the problem of high inference latency in generative flow and diffusion models for robotic policies by proposing the One-Step Flow Policy (OFP), which achieved state-of-the-art results across 56 simulated manipulation tasks, outperforming 100-step models while accelerating action generation by over 100x.
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 $Ï_{0.5}$ 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.