分层控制架构让人形机器人更稳地走楼梯和台阶。
Architecture Is All You Need: Diversity-Enabled Sweet Spots for Robust Humanoid Locomotion
- 高低时序分离：高速本体反馈+低速感知决策，分工明确。
- 仿真与实机均胜过端到端模型，在台阶任务上成功率达100%。
- 无需复杂网络，简单结构也能实现鲁棒运动，适合硬件部署。
在非结构化环境中实现人形机器人稳定行走，需要在快速底层稳定与慢速感知决策之间取得平衡。本文提出一种简单的分层控制架构（LCA），即高频运行的本体感觉稳定器，搭配低频紧凑的感知策略，即使使用极简感知编码器，也显著优于单一的端到端设计。通过两阶段训练流程（先盲稳定器预训练，再感知微调），分层策略在仿真和真实硬件上均持续超越单阶段方案。在Unitree G1人形机器人上，该方法成功完成楼梯和台阶任务，而单阶段感知策略则失败。结果表明，时间尺度的架构分离，而非网络规模或复杂度，是实现鲁棒感知驱动行走的关键。
原文摘要 · Abstract (English)
Robust humanoid locomotion in unstructured environments requires architectures that balance fast low-level stabilization with slower perceptual decision-making. We show that a simple layered control architecture (LCA), a proprioceptive stabilizer running at high rate, coupled with a compact low-rate perceptual policy, enables substantially more robust performance than monolithic end-to-end designs, even when using minimal perception encoders. Through a two-stage training curriculum (blind stabilizer pretraining followed by perceptual fine-tuning), we demonstrate that layered policies consistently outperform one-stage alternatives in both simulation and hardware. On a Unitree G1 humanoid, our approach succeeds across stair and ledge tasks where one-stage perceptual policies fail. These results highlight that architectural separation of timescales, rather than network scale or complexity, is the key enabler for robust perception-conditioned locomotion.
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。