发表机构
University of Florida（佛罗里达大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
针对屋顶施工中人形机器人在倾斜表面的运动挑战，提出任务语义场景锚定框架，结合轨迹优化与执行感知强化学习，在Unitree G1上实现满足支撑、间隙和非穿透标准的全身运动，误差低于0.531厘米。
AI 中文摘要
屋顶施工要求工人在倾斜表面上协调运动、平衡以及与工作相关的身体动作，这对人形机器人而言是一项具有挑战性的应用。然而，直接重定向的人类演示可能保留了运动外观，却使机器人的脚或手相对于屋顶的位置不正确。本研究提出了一种任务语义场景锚定框架，用于在 Unitree G1 上学习屋顶工人风格的全身运动。使用跟踪系统捕捉人类演示并将其重定向到机器人，而度量屋顶模型则提供了跟踪系统无法获取的空间参考。轨迹级优化将推断出的支撑接触点和标注的工作关系锚定到屋顶上，执行感知的强化学习则鼓励所得策略在动态跟踪误差下保持这些关系。该框架通过多运动跟踪研究、屋顶坡度覆盖矩阵、五向射钉枪消融实验、锤击和侧向推挤的跨任务实验，以及与纯强化学习和零样本遥操作的比较进行评估。我们的方法使机器人能够在所有评估种子中满足支撑、工作间隙和非穿透标准。在射钉枪、锤击和推挤任务中，该方法实现了 0.256 至 0.531 厘米的工作间隙误差，且每个任务均获得 3/3 的成功评估。物理实验复现了上坡行走、射钉枪、锤击和弯腰动作，平均基座框架运动误差低于 80 毫米。这些发现确立了场景锚定的人类运动学习作为面向施工的人形机器人运动原语的有前景基础。
英文摘要
Roofing requires workers to coordinate locomotion, balance, and work-related body motions on pitched surfaces, creating a challenging application for humanoid robots. Directly retargeted human demonstrations, however, may preserve motion appearance while placing the robot's feet or hands incorrectly relative to the roof. This study presents a task-semantic scene-grounded framework for learning roofer-style whole-body motions on a Unitree G1. Human demonstrations are captured using a tracking system and retargeted to the robot, while a metric roof model supplies the spatial reference unavailable from the tracking system. A trajectory-level optimization grounds inferred support contacts and annotated work relations to the roof, and execution-aware reinforcement learning encourages the resulting policy to preserve these relations under dynamic tracking errors. The framework is evaluated through a multi-motion tracking study, a roof-pitch coverage matrix, a five-way nailgun ablation, cross-task experiments on hammering and lateral pushing, and comparisons with pure reinforcement learning and zero-shot teleoperation. Our method enables the robot to satisfy support, work-clearance, and nonpenetration criteria across all evaluated seeds. Across nailgun, hammering, and pushing, it achieves work-clearance errors between 0.256 and 0.531 cm and 3/3 successful evaluations per task. Physical experiments reproduce uphill walking, nailgun, hammering, and bending motions with mean base-frame motion errors below 80 mm. These findings establish scene-grounded human motion learning as a promising basis for construction-oriented humanoid motion primitives.
