# ImpedanceMimic：学习阻抗动作空间实现零样本力觉运动模仿

> 原文链接：https://robotics.hellobike.com/papers/2026-11-17-2611-19153-94bca482
> 频道：papers | 更新日期：2026-11-17 | 标签：auto, arxiv
> 核心摘要：提出ImpedanceMimic强化学习策略，通过调节关节刚度与阻尼解决固定增益PD控制中运动与力跟踪的冲突。在129人数据集上验证其力觉保真度优于传统控制器，并实现向未见人类及人形机器人的零样本迁移。

# ImpedanceMimic：学习阻抗动作空间实现零样本力觉运动模仿

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[ImpedanceMimic: Learned Impedance as Action Space for Force-Accurate Motion Imitation](https://arxiv.org/abs/2611.19153)（arXiv 2611.19153）

## 核心内容

针对运动模仿中接触力恢复的病态问题，研究提出ImpedanceMimic框架，将每关节刚度、阻尼及位置设定点作为动作空间进行端到端学习。该方法仅需运动数据即可推理，在包含129名参与者、6种活动的基准测试中，其运动学与力觉保真度全面超越匹配固定增益PD、直接转矩及肌肉驱动控制器，性能逼近专用运动-力回归器，有效解决了摆动相与接触相对关节阻抗的冲突需求。

技术上通过将增益表示为惯性缩放先验的倍数，使单一策略无需显式人体测量输入即可适配特定生物力学模型，并成功零样本迁移至未见过的人类受试者及人形机器人本体。学习到的阻抗构成了跨身体共享的低维运动状态表征，展现出符合生物力学规律的泛化特性：刚度随负载自动缩放，阻尼则能预判卸载过程，为具身智能体的高保真力觉模仿提供了通用控制范式。

## 关键看点

- **动态阻抗调节**：RL策略实时输出每关节刚度、阻尼与位置设定点，仅依赖运动数据推理，解决固定增益PD在接触与摆动相的性能权衡难题
- **零样本跨体迁移**：采用惯性缩放先验归一化增益，单一策略无需人体测量参数即可适配不同生物力学模型，并直接迁移至人形机器人
- **力觉保真度验证**：在129人6类活动数据集上，运动与力跟踪精度优于直接转矩及现有模仿控制器，逼近专用力回归器上限
- **低维状态表征**：学习到的阻抗具备物理可解释性，刚度随负载线性缩放且阻尼预判卸载，形成跨个体共享的通用运动状态编码

## 工程与产业观察

该工作证明了将可变阻抗作为RL动作空间是解决具身模仿学习中“形似神不似”问题的关键路径。通过惯性归一化实现的零样本跨体迁移，大幅降低了人形机器人获取高质量力觉示范数据的门槛，为从人类演示到机器人本体的力控技能复用提供了可落地的算法基座。

## 英文摘要 (Abstract)

Recovering the forces behind a motion is ill-posed: different controllers can reproduce the same movement while generating different contact forces. Physics-based motion imitation resolves this ambiguity through simulation, but relies on fixed-gain proportional-derivative (PD) control. We show that fixed gains impose a trade-off between motion and force tracking, as contact and swing place conflicting demands on joint stiffness and damping. We introduce ImpedanceMimic, a reinforcement-learning policy that modulates per-joint stiffness, damping, and position setpoints from motion and force measurements, but requires only motion at inference. Across 129 participants and six activities, ImpedanceMimic outperforms matched fixed-gain PD under every integration scheme, direct-torque, muscle-actuated and existing imitation controllers in kinematic and force fidelity, approaching a dedicated motion-to-force regressor. By expressing gains as multiples of inertia-scaled priors, a single policy controls subject-specific biomechanical models without explicit anthropometric input and transfers zero-shot to unseen humans and a humanoid robot. The learned impedance forms a low-dimensional representation of motor state, shared across bodies, that generalizes to withheld activities: stiffness scales with load, and damping anticipates unloading.

作者：Matthias Jammot, Pembe Gizem \"Ozdil
:::

## 原始参考来源 (Sources)

- [arXiv 2611.19153：ImpedanceMimic: Learned Impedance as Action Space for Force-Accurate Motion Imitation](https://arxiv.org/abs/2611.19153)