Skip to content

论文arXiv 2303.041372023-03-07

Diffusion Policy:视运动策略学习新范式,12项任务平均性能提升46.9% ​

本文由 AI 根据公开内容整理,可能有误,请以原文为准:Diffusion Policy: Visuomotor Policy Learning via Action Diffusion(arXiv 2303.04137)

核心内容 ​

哥伦比亚大学等机构提出Diffusion Policy,将机器人行为生成重构为条件去噪扩散过程。该方法在4个主流机器人操作基准的12项任务中进行评测,相比现有最先进机器人学习方法,取得了46.9%的平均性能提升。研究验证了扩散模型在物理机器人视运动策略学习中的有效性,并公开了完整代码、数据集及训练细节,为具身智能策略学习提供了新的生成式建模范式。

技术上,该策略通过学习动作分布得分函数的梯度,并在推理阶段经由随机Langevin动力学迭代优化。针对机器人控制特性,论文引入了滑动时域控制(Receding Horizon Control)、视觉条件注入及时间序列扩散Transformer三大关键改进。这种架构不仅天然适配高维动作空间,还能优雅处理多模态动作分布,且在训练过程中表现出极高的稳定性,克服了传统行为克隆方法的局限。

关键看点 ​

  • 性能基准:在4大机器人操作基准共12项任务中,较现有SOTA方法实现46.9%的平均性能提升
  • 核心机制:基于动作分布得分函数梯度学习,推理时通过随机Langevin动力学步骤进行迭代优化
  • 关键技术:集成滑动时域控制、视觉条件化与时间序列扩散Transformer,适配物理机器人控制
  • 工程优势:原生支持多模态动作分布与高维动作空间,训练稳定性显著优于传统策略学习方法

工程与产业观察 ​

Diffusion Policy确立了生成式模型在机器人底层控制中的工程可行性。其通过滑动时域控制与时序Transformer解决了扩散模型推理慢、难对齐物理约束的痛点,46.9%的性能跃升证明了从判别式向生成式策略迁移的价值。开源资产为具身智能社区提供了高价值基线,加速了复杂操作任务的算法迭代。

英文摘要 (Abstract) ​

This paper introduces Diffusion Policy, a new way of generating robot behavior by representing a robot's visuomotor policy as a conditional denoising diffusion process. We benchmark Diffusion Policy across 12 different tasks from 4 different robot manipulation benchmarks and find that it consistently outperforms existing state-of-the-art robot learning methods with an average improvement of 46.9%. Diffusion Policy learns the gradient of the action-distribution score function and iteratively optimizes with respect to this gradient field during inference via a series of stochastic Langevin dynamics steps. We find that the diffusion formulation yields powerful advantages when used for robot policies, including gracefully handling multimodal action distributions, being suitable for high-dimensional action spaces, and exhibiting impressive training stability. To fully unlock the potential of diffusion models for visuomotor policy learning on physical robots, this paper presents a set of key technical contributions including the incorporation of receding horizon control, visual conditioning, and the time-series diffusion transformer. We hope this work will help motivate a new generation of policy learning techniques that are able to leverage the powerful generative modeling capabilities of diffusion models. Code, data, and training details is publicly available diffusion-policy.cs.columbia.edu

作者:Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burchfiel, Russ Tedrake, Shuran Song

来源

本页事实对应的公开页面,以原文为准。

  1. arXiv 2303.04137:Diffusion Policy: Visuomotor Policy Learning via Action Diffusion访问 2026-10-23