适用于移动机械手无关节状态反馈情况的基于人-机-机协作的无标定视觉伺服控制

发布时间：2018-10-23 20:27

【摘要】：移动机器人在执行某些危险品处置任务时,其搭载的机械臂的关节状态可能会因为测量器件的损坏而无法获知.为了不在召回机器人上延误危险品的处置时间,提出一种适用于移动机械手无关节状态反馈情况的基于人-机-机协作(HRRC)的无标定视觉伺服控制系统.首先建立能反映机械手关节状态的虚拟外骨骼(虚拟模型),方法是在另一台机器人在线拍摄到的监控图像上,使用人机交互(HCI)输入设备(如鼠标)手动框选关节所在的区域.虚拟外骨骼与多关节跟踪算法配合可以实现对机械臂对应关节的导向控制及末端姿态保持.为了利用人工导引点对虚拟外骨骼进行导引控制,通过广义回归神经网络(GRNN)来映射虚拟外骨骼的末端与关节角的关系.最后,在轴孔装配实验中,本文方法能够使机械手末端在人工导引下完成任务,且能使末端姿态误差保持在±1?左右;而常规的单关节控制方式无法控制机械手末端完成任务,且无法实现末端姿态保持.对比实验结果验证了本文提出的控制系统在无关节状态反馈的情况下能够帮助操作人员直观地使用机械手处置目标,并且在此过程中能够使机械手末端保持在指定姿态.
[Abstract]:When the mobile robot performs some dangerous goods disposal tasks, the joint state of the robot arm may not be known because of the damage of the measuring device. In order not to delay the disposal time of dangerous goods on the recalled robot, an uncalibrated visual servo control system based on human-machine cooperative (HRRC) is proposed for mobile manipulators without joint state feedback. Firstly, a virtual exoskeleton (virtual model) which can reflect the joint state of the manipulator is established. The method is to select the joint area by using the manual frame of the human-computer interactive (HCI) input device (such as mouse) on the monitoring image captured online by another robot. Virtual exoskeleton combined with multi-joint tracking algorithm can realize the steering control and terminal attitude maintenance of the corresponding joints of the manipulator. In order to control virtual exoskeleton by artificial guidance points, the relationship between the end of virtual exoskeleton and joint angle is mapped by generalized regression neural network (GRNN). Finally, in the experiment of shaft hole assembly, the method in this paper can make the end of manipulator complete the task under manual guidance, and can keep the error of terminal attitude in 卤1? The conventional single joint control method can not control the end of the manipulator to complete the task, and can not achieve terminal attitude maintenance. The experimental results show that the proposed control system can help the manipulator to handle the target intuitively without joint state feedback, and can keep the manipulator terminal in a specified position during the process.
【作者单位】：东南大学仪器科学与工程学院;
【基金】：国家自然科学基金(61325018,61272379)
【分类号】：TP242

【相似文献】