让机器人在没训练过的情况下，自动避开自身碰撞完成任务。
EmbodiSteer: Steering Embodiment-Agnostic Visuomotor Policies with Joint-Space Guidance for Zero-Shot Cross-Embodiment Deployment
- 用关节空间引导扩散模型，让无体感策略学会避障。
- 仿真中碰撞率降46.1%，成功率升28.5%；实机降90%碰撞。
- 无需重新训练，可直接部署到不同机器人上使用。
大规模机器人模仿学习依赖来自多种机器人或无体感数据的异构数据，使笛卡尔空间末端执行器动作成为无体感策略学习的关键接口。然而，仅基于末端执行器的抽象会使笛卡尔策略忽略部署机器人的本体结构，在全身碰撞规避等机器人特异性约束下变得脆弱。为此，我们提出EmbodiSteer，一个无需训练的框架，实现无体感视觉运动策略的零样本、本体感知部署。EmbodiSteer在笛卡尔空间保持策略学习的同时，通过正向运动学和基于雅可比的更新，将推理时的扩散采样高效映射到目标机器人的关节空间。每次去噪步骤后，通过全身体碰撞感知引导关节轨迹，使机械臂能避开碰撞同时保留学习到的末端行为。相比仅笛卡尔执行，EmbodiSteer在9个模拟机器人上将碰撞率降低46.1%，任务成功率提升28.5%；在两个物理机器人高度受限场景中，碰撞率进一步降低90.0%，成功率提高36.7%。
原文摘要 · Abstract (English)
Scalable robot imitation learning relies on large-scale heterogeneous data from diverse robots or body-free data, making Cartesian end-effector actions a key interface for embodiment-agnostic policy learning. However, end-effector-only abstraction leaves Cartesian policies unaware of the deployed robot body, making them brittle under robot-specific constraints such as whole-body collision avoidance. To overcome this limitation, we present EmbodiSteer, a training-free framework that steers embodiment-agnostic visuomotor policies toward zero-shot, embodiment-aware deployment. EmbodiSteer keeps policy learning in Cartesian space while efficiently lifting inference-time diffusion sampling into the target robot's joint space via forward kinematics and Jacobian-based updates. With whole-body collision-aware guidance over joint trajectories after each denoising step, the arm can be steered away from collisions while preserving learned end-effector behavior. Compared with Cartesian-only execution, EmbodiSteer reduces collision rate by 46.1% and improves task success rate by 28.5% across 9 simulated robots, and further achieves 90.0% collision rate reduction and 36.7% success rate increase on two physical robots in highly constrained scenarios. Our project page is at https://frankwang67.github.io/EmbodiSteer-Page.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。