发表机构
The Robotics Institute, School of Computer Science, Carnegie Mellon University; Department of Aerospace Engineering, Pennsylvania State University; School of Mechanical Engineering, Kyung Hee University（卡内基梅隆大学计算机学院机器人研究所; 宾夕法尼亚州立大学航空航天工程系; 庆熙大学机械工程学院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究提出AM-Bench模块化仿真套件与基准，用于多旋翼空中操作策略学习，可系统评估实体、控制等因素的相互作用，经仿真与硬件测试验证其诊断价值。
AI 中文摘要
标准化基准在推动机器人操作学习发展中发挥了核心作用，但多数聚焦于地面支撑的操作系统，这限制了其在动态至关重要的领域（如空中操作AM）的适用性。AM面临独特的系统级挑战，包括环境干扰、操作器与浮动基座间的耦合动力学以及受限的自由度，因此任务性能取决于机器人实体、低层控制与高层策略设计的共同作用。我们提出AM-Bench，这是一款面向多旋翼AM策略学习的模块化仿真套件与基准，涵盖欠驱动、全驱动与过驱动等代表性实体，包含接触、运输与约束交互类共12项任务，可配置气动干扰与执行器饱和、提供标准低层控制器及基线策略学习算法。与主要强调端到端策略性能的现有操作基准不同，AM-Bench可对实体、控制、干扰及策略选择的相互作用进行系统级评估。我们通过三项仿真研究（涵盖高层策略、策略-控制接口及实体）展示其诊断价值，并对建模效果进行了实际验证，同时完成了学习流程的硬件测试。
英文摘要
Standardized benchmarks have played a central role in advancing robot manipulation learning, yet most focus on ground-supported manipulation systems, which limits their applicability to dynamics-critical domains such as aerial manipulation (AM). AM presents distinct system-level challenges, including environmental disturbances, coupled dynamics between the manipulator and floating base, and constrained degrees of freedom. Consequently, task performance depends jointly on robot embodiment, low-level control, and high-level policy design. We introduce AM-Bench, a modular simulation suite and benchmark for multirotor-based AM policy learning. AM-Bench includes representative embodiments spanning underactuated, fully actuated, and overactuated systems, 12 tasks across contact, transport, and constrained interaction, configurable aerodynamic disturbances and actuator saturation, standard low-level controllers, and baseline policy-learning algorithms. Unlike prior manipulation benchmarks that primarily emphasize end-to-end policy performance, AM-Bench enables system-level evaluation of how embodiment, control, disturbances, and policy choices interact. We demonstrate its diagnostic value through three simulation studies spanning high-level policies, policy--control interfaces, and embodiments, together with real-world validation of modeled effects and a hardware test of the learning pipeline.
Comments28 pages, 7 figures, 15 tables
