arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2511.14756cs.RO
收藏

HMC:学习接触密集型移动操作的异构元控制

HMC: Learning Heterogeneous Meta-Control for Contact-Rich Loco-Manipulation

UC San Diego(加州大学圣地亚哥分校)

Lai Wei, Xuanbin Peng, Ri-Zhao Qiu, Tianshu Huang, Xuxin Cheng, Xiaolong Wang

AI总结:

针对纯位置控制器在接触或变负载下的不足,提出异构元控制(HMC)框架,通过HMC-Controller融合多控制模态、HMC-Policy构建异构架构,在真实人形机器人实验中任务性能相对提升超50%。

摘要:

从真实世界机器人演示中学习有望实现与复杂真实环境的交互。然而,交互动力学的复杂性和可变性常导致纯位置控制器在处理接触或变负载时面临困难。为此,我们提出用于移动操作的异构元控制(HMC)框架,自适应拼接位置、阻抗和混合力-位置三种控制模态。首先引入HMC-Controller接口,在扭矩空间连续融合不同控制配置文件的动作,同时支持遥操作与策略部署。然后,为学习鲁棒的力感知策略,提出HMC-Policy将不同控制器统一为异构架构,采用专家混合式路由从大规模纯位置数据和细粒度力感知演示中学习。在真实人形机器人上的实验显示,在柔顺擦桌、抽屉开启等挑战性任务中,相比基线方法相对提升超50%,证明了HMC的有效性。

英文摘要:

Learning from real-world robot demonstrations holds promise for interacting with complex real-world environments. However, the complexity and variability of interaction dynamics often cause purely positional controllers to struggle with contacts or varying payloads. To address this, we propose a Heterogeneous Meta-Control (HMC) framework for Loco-Manipulation that adaptively stitches multiple control modalities: position, impedance, and hybrid force-position. We first introduce an interface, HMC-Controller, for blending actions from different control profiles continuously in the torque space. HMC-Controller facilitates both teleoperation and policy deployment. Then, to learn a robust force-aware policy, we propose HMC-Policy to unify different controllers into a heterogeneous architecture. We adopt a mixture-of-experts style routing to learn from large-scale position-only data and fine-grained force-aware demonstrations. Experiments on a real humanoid robot show over 50% relative improvement vs. baselines on challenging tasks such as compliant table wiping and drawer opening, demonstrating the efficacy of HMC.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑