引用本文:陈彦杰,兰立民,吴杨宁,张辉,王耀南.飞行作业机器人的视觉伺服跟踪与物理交互导纳控制[J].控制理论与应用,2025,42(7):1367~1378.[点击复制]
CHEN Yan-jie,LAN Li-min,WU Yang-ning,ZHANG Hui,WANG Yao-nan.Visual-servoing tracking and physical interaction admittance control for an unmanned aerial manipulator[J].Control Theory & Applications,2025,42(7):1367~1378.[点击复制]
飞行作业机器人的视觉伺服跟踪与物理交互导纳控制
Visual-servoing tracking and physical interaction admittance control for an unmanned aerial manipulator
摘要点击 3002  全文点击 309  投稿时间：2023-07-23  修订日期：2025-05-09
查看全文  查看/发表评论  下载PDF阅读器   HTML
DOI编号  10.7641/CTA.2024.30498
  2025,42(7):1367-1378
中文关键词  飞行作业机器人  视觉伺服  非奇异快速终端滑模控制  反演控制  自适应控制
英文关键词  unmanned aerial manipulator  visual servoing  non-singular fast terminal sliding mode control  backstepping control  adaptive control
基金项目  科技创新2030–“新一代人工智能”重大课题项目(2021ZD0114503), 国家自然科学基金项目(62273098, 61971071, 62027810, 62133005), 国家重大 研究计划–重点支持项目(92148204)资助.
作者单位E-mail
陈彦杰 福州大学机械工程及自动化学院 chenyanjie@hnu.edu.cn 
兰立民 福州大学机械工程及自动化学院  
吴杨宁 福州大学机械工程及自动化学院  
张辉* 湖南大学机器人学院  
王耀南 湖南大学电气与信息工程学院  
中文摘要
      飞行作业机器人在执行动态目标跟踪及物理交互作业时, 通常存在速度信息难获取、跟踪控制速度慢和物 理交互扰动强等问题. 本文首先针对速度信息难获取问题, 设计了一种基于图像动力学的自适应速度观测器, 通过 估计飞行作业机器人与动态目标之间的相对速度, 降低了速度信息测量的噪声与误差; 然后, 考虑跟踪控制速度慢 问题, 设计了有限时间内收敛的非奇异快速终端滑模视觉伺服控制方法, 实现了动态目标的稳定跟踪; 接着, 考虑物 理交互作业时机载机械臂对飞行作业机器人造成的强扰动问题, 提出了一种具有自适应上界估计能力的机械臂反 演导纳控制策略, 在降低力传感器噪声影响的基础上, 实现了持续物理交互时的柔顺控制性能; 最后, 在仿真平台与 真实实验场景中设计动态目标跟踪及物理交互作业实验. 结果表明, 所提出方法可在无动态目标相对速度信息的 情况下实现快速视觉伺服跟踪效果, 并在与人发生物理交互时实现稳定的控制性能.
英文摘要
      When engaging in tasks involving dynamic target tracking and physical interactions, unmanned aerial manipulator (UAM) commonly encounter challenges, such as the difficulty of obtaining speed information, slow tracking control speeds and strong disturbances during physical interactions. This paper presents a series of solutions to address these challenges. Firstly, an image-based adaptive velocity observer is introduced to mitigate the issue of acquiring speed information. By estimating the relative velocity between the UAM and dynamic targets, this approach reduces noise and errors in speed measurements. Subsequently, to address the problem of slow tracking control speeds, a non-singular fast terminal sliding mode visual servo control method is proposed, ensuring stable tracking of dynamic targets within a finite time frame. Moreover, considering the significant perturbations introduced by onboard manipulators during physical contact operations, a backstepping admittance control strategy with adaptive upper-bound estimation is proposed for the onboard manipulator to achieve compliant behavior while minimizing the influence of force sensor noise. Finally, the effectiveness of our approach is validated through dynamic target tracking and physical interaction experiments conducted in both simulated and real-world scenarios. The results demonstrate the capability to achieve rapid visual servoing tracking performance for UAMs without relying on relative velocity information of dynamic targets and the ability to maintain stable control performance during physical interactions with humans.