论文arXiv 2611.01729今天

ODE Policy：基于动作动力学建模实现机器人策略连续生成与高效泛化

提出ODE Policy通过学习动作向量场替代离散动作映射，在IL任务中成功率达Flow Matching的5.9倍、推理快10.8倍。该架构作为VLA/WAM动作头及RL Actor，在单双臂、灵巧手及人形机器人等14项真实任务中取得最高成功率。

AI 整理

本体 SDK 与控制
具身模型

本文由 AI 根据公开内容整理，可能有误，请以原文为准：ODE Policy: Robot Learning via Action Dynamics, Not Actions（arXiv 2611.01729）

核心内容

针对传统机器人策略将物理运动连续性隐式学习的问题，研究提出ODE Policy，利用神经常微分方程直接学习动作向量场而非离散动作值。该方法通过匹配短时动作块或单步控制目标来积分动作，并引入轻量级多样性头处理多模态行为。评估显示，其在模仿学习中仅训练一个epoch，成功率、泛化能力和模式平衡性分别达到Flow Matching基线的5.9倍、5.4倍和3.9倍，且动作生成速度提升10.8倍，轨迹平滑度优于人类示教数据。

ODE Policy具有高度通用性，可作为视觉运动策略、VLA与世界模型的动作头，以及Model-based RL的Actor。在涵盖单臂、双臂、灵巧操作、移动底盘及人形机器人的14项真实世界任务测试中，该方法均实现了最高成功率。在强化学习场景下，它还能加速SuperMimic算法的人形机器人运动跟踪收敛并平滑输出动作。相比现有基线，ODE Policy在学习效率、数据利用率、参数量及执行效率上均展现出显著优势，为规模化机器人学习提供了新路径。

关键看点

核心机制：采用神经ODE学习动作向量场，部署时从上一时刻积分生成动作，从构造上保证执行轨迹的物理连续性
IL性能跃升：对比Flow Matching基线，训练1 epoch后成功率提升5.9倍、泛化能力提升5.4倍、推理速度快10.8倍
广泛适配性：兼容VLA、WAM及Model-based RL架构，在单双臂、灵巧手及人形机器人等14项真机任务中获最高成功率
多模态处理：集成轻量级多样性头（Diversity Head），通过随机化向量场方向有效捕获复杂操作中的多模态行为分布

工程与产业观察

ODE Policy将动作生成从离散回归转变为连续动力学建模，从根本上解决了高频控制下的轨迹抖动问题。其10.8倍的推理加速与5.9倍的成功率提升，证明了显式物理归纳偏置在具身智能中的关键价值。该架构对VLA和人形机器人控制的高效适配，为端侧实时部署与大规模真机训练提供了兼顾平滑性与计算效率的工程范式。

英文摘要 (Abstract)

Most robot policies, whether trained by imitation learning (IL) or reinforcement learning (RL), map observations to action values at discrete control steps and leave the continuity of physical motion to be learned implicitly. We propose ODE policy, which learns the dynamics of the action rather than the action itself. A neural ordinary differential equation (ODE) learns the vector field of the action by matching the integrated actions to the targets over short action chunks in IL and single control steps in RL. A lightweight diversity head captures multi-modal behavior by randomizing the direction of the field. At deployment, each action is integrated from the previous one, so the executed trajectory is continuous by construction. We evaluate it as a visuomotor policy, as the action head of vision-language-action (VLA) and world action models (WAMs), and as the actor of model-based RL against 14 baselines on unimanual, bimanual, dexterous, locomotion, and loco-manipulation tasks. In IL, against flow matching, it reaches 5.9$\times$ the success, 5.4$\times$ the generalization, and 3.9$\times$ the mode balance after one epoch, and generates actions 10.8$\times$ faster. It mostly moves more smoothly than human demonstrations. On 14 real-world tasks across single-arm, dual-arm, dexterous, and humanoid robots, it achieves the highest success rate. In RL, it further speeds up SuperMimic for humanoid motion tracking and smooths its actions. Efficient in learning, data, parameters, and execution, ODE policy offers a promising path to scaling robot learning.

作者：Jindou Jia, Jingliang Li, Kuangji Zuo, Tuo An, Yuxuan Hu, Mengfei Zhao, Bofan Lyu, Jiaqi Bai, Daoyuan Zhu, Hengyu Dai, Junpeng Zhang, Jianfei Yang