外观
论文arXiv 2509.180462 个月前
HuMam：通过使用 Mamba 的端到端深度强化学习实现人形运动控制
原文链接：HuMam: Humanoid Motion Control via End-to-End Deep Reinforcement Learning with Mamba（arXiv 2509.18046 · 发布日期：2026-08-01）
HuMam: Humanoid Motion Control via End-to-End Deep Reinforcement Learning with Mamba
HuMam面向人形运动端到端RL不稳定和能耗高问题，采用单层Mamba融合状态与足步目标，在mc-mujoco的JVRC-1上验证。
核心要点
- 端到端人形运动RL映射紧凑，但常受训练不稳定、特征融合低效和驱动成本高影响。
- 方法以机器人状态为中心，用单层Mamba编码器融合机器人状态、定向足步目标和连续相位时钟。
- 策略输出关节位置目标由低层PD跟踪，并通过PPO与六项奖励进行优化。
- 在JVRC-1的mc-mujoco实验中，相比强前馈基线提升学习效率、稳定性和任务表现，并降低功率与力矩峰值。
验证范围与局限：验证范围为mc-mujoco中的JVRC-1人形机器人。
英文摘要 (Abstract)
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.
作者：Yinuo Wang, Xiaowen Tao, Yunfeng Hu, Yang Liang, Ziyu Song, Haitao Ding, Jinzhao Zhou
来源
本页事实对应的公开页面，以原文为准。