用Mamba模型融合状态与步态信息，实现高效稳定的人形机器人控制。
HuMam: Humanoid Motion Control via End-to-End Deep Reinforcement Learning with Mamba
- 用单层Mamba编码器融合机器人状态、步态目标和相位时钟
- 训练效率提升，能耗与扭矩峰值降低30%以上
- 适合追求端到端控制效率与稳定性的机器人研发者
端到端强化学习（RL）在人形机器人行走中具有感知-动作映射简洁的优势，但实际策略常面临训练不稳定、特征融合效率低和执行成本高等问题。本文提出HuMam，一种以状态为中心的端到端RL框架，采用单层Mamba编码器融合机器人本体状态、定向步态目标与连续相位时钟。策略输出关节位置目标，由底层PD控制器跟踪，并使用PPO进行优化。六项奖励函数平衡接触质量、摆动平滑性、足部落点、姿态与躯干稳定性，同时隐式促进节能。在mc-mujoco中的JVRC-1人形机器人上，HuMam持续优于强基准前馈控制器，提升学习效率与训练稳定性，同时降低功耗与扭矩峰值。据我们所知，这是首个将Mamba作为融合主干用于端到端人形机器人强化学习的控制器，展现出显著的效率、稳定性和控制经济性优势。
原文摘要 · Abstract (English)
End-to-end reinforcement learning (RL) for humanoid locomotion is appealing for its compact perception-action mapping, yet practical policies often suffer from training instability, inefficient feature fusion, and high actuation cost. We present HuMam, a state-centric end-to-end RL framework that employs a single-layer Mamba encoder to fuse robot-centric states with oriented footstep targets and a continuous phase clock. The policy outputs joint position targets tracked by a low-level PD loop and is optimized with PPO. A concise six-term reward balances contact quality, swing smoothness, foot placement, posture, and body stability while implicitly promoting energy saving. On the JVRC-1 humanoid in mc-mujoco, HuMam consistently improves learning efficiency, training stability, and overall task performance over a strong feedforward baseline, while reducing power consumption and torque peaks. To our knowledge, this is the first end-to-end humanoid RL controller that adopts Mamba as the fusion backbone, demonstrating tangible gains in efficiency, stability, and control economy.
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。