# ALMI框架实现人形机器人上下身对抗策略学习与全身协调控制

> 原文链接：https://robotics.hellobike.com/papers/2025-04-19-2504-14305-bbfd345d
> 频道：papers | 更新日期：2025-04-19 | 标签：auto, arxiv, cs.RO
> 核心摘要：针对人形机器人全身协调难题，提出ALMI对抗学习框架，解耦上下身策略并在Unitree H1实机验证。同步开源MuJoCo高质量全身运动数据集，支持从仿真到真机的稳定部署与遥操作扩展。

# ALMI框架实现人形机器人上下身对抗策略学习与全身协调控制

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning](https://arxiv.org/abs/2504.14305)（arXiv 2504.14305）

## 核心内容

针对人形机器人模仿人类动作时因忽视上下身角色差异导致的策略学习算力高、实机易跌倒问题，研究团队提出Adversarial Locomotion and Motion Imitation（ALMI）框架。该方法通过对抗机制解耦控制，在仿真及全尺寸Unitree H1机器人上均实现了鲁棒移动与精准动作跟踪，并发布了包含高质量MuJoCo轨迹的开源全身运动控制数据集，项目主页已上线供社区复现。

ALMI核心在于构建上下身对抗策略：下肢策略专注速度指令跟随以保障移动稳定性，上身策略负责动作跟踪，两者通过迭代更新实现协同，避免传统端到端方法的耦合失稳。该架构可无缝扩展至遥操作loco-manipulation任务。配套开源数据集提供可直接部署于真机的episodic trajectories，解决了具身智能研究中高保真全身运动数据稀缺的工程痛点，显著降低Sim-to-Real迁移门槛。

## 关键看点

- **对抗解耦架构**：下身策略追踪速度指令保稳定，上身策略跟踪动作，迭代更新实现全身协调控制
- **实机验证平台**：在全尺寸Unitree H1机器人完成部署，证实仿真训练策略具备Sim-to-Real迁移能力
- **开源数据资产**：发布基于MuJoCo的高质量全身运动控制数据集，含可直接用于真机部署的轨迹片段
- **任务扩展能力**：框架支持接入遥操作系统，将全身协调控制能力延伸至loco-manipulation复合任务

## 工程与产业观察

ALMI通过对抗解耦降低了人形机器人全身策略学习的维度灾难风险，为H1等本体提供了稳定的运动基线。其开源的高保真MuJoCo轨迹数据集填补了具身运动数据缺口，对算法工程师优化Sim-to-Real流程及产业界构建通用运动原语具有重要工程参考价值。

## 英文摘要 (Abstract)

Humans exhibit diverse and expressive whole-body movements. However, attaining human-like whole-body coordination in humanoid robots remains challenging, as conventional approaches that mimic whole-body motions often neglect the distinct roles of upper and lower body. This oversight leads to computationally intensive policy learning and frequently causes robot instability and falls during real-world execution. To address these issues, we propose Adversarial Locomotion and Motion Imitation (ALMI), a novel framework that enables adversarial policy learning between upper and lower body. Specifically, the lower body aims to provide robust locomotion capabilities to follow velocity commands while the upper body tracks various motions. Conversely, the upper-body policy ensures effective motion tracking when the robot executes velocity-based movements. Through iterative updates, these policies achieve coordinated whole-body control, which can be extended to loco-manipulation tasks with teleoperation systems. Extensive experiments demonstrate that our method achieves robust locomotion and precise motion tracking in both simulation and on the full-size Unitree H1 robot. Additionally, we release a large-scale whole-body motion control dataset featuring high-quality episodic trajectories from MuJoCo simulations deployable on real robots. The project page is https://almi-humanoid.github.io.

作者：Jiyuan Shi, Xinzhe Liu, Dewei Wang, Ouyang Lu, Sören Schwertfeger, Chi Zhang, Fuchun Sun, Chenjia Bai, Xuelong Li
:::

## 原始参考来源 (Sources)

- [arXiv 2504.14305：Adversarial Locomotion and Motion Imitation for Humanoid Policy Learning](https://arxiv.org/abs/2504.14305)