纪鹏, 宋爱国, 吴常铖, 魏宏明, 宋子墨, 张琪. 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[J]. 机器人, 2017, 39(2): 197-204. DOI: 10.13973/j.cnki.robot.2017.0197
引用本文: 纪鹏, 宋爱国, 吴常铖, 魏宏明, 宋子墨, 张琪. 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[J]. 机器人, 2017, 39(2): 197-204. DOI: 10.13973/j.cnki.robot.2017.0197
JI Peng, SONG Aiguo, WU Changcheng, WEI Hongming, SONG Zimo, ZHANG Qi. Human-Robot-Robot-Cooperation Based Uncalibrated Visual Servoing Control for Mobile Robotic Manipulators without Joint-State Feedback[J]. ROBOT, 2017, 39(2): 197-204. DOI: 10.13973/j.cnki.robot.2017.0197
Citation: JI Peng, SONG Aiguo, WU Changcheng, WEI Hongming, SONG Zimo, ZHANG Qi. Human-Robot-Robot-Cooperation Based Uncalibrated Visual Servoing Control for Mobile Robotic Manipulators without Joint-State Feedback[J]. ROBOT, 2017, 39(2): 197-204. DOI: 10.13973/j.cnki.robot.2017.0197

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

Human-Robot-Robot-Cooperation Based Uncalibrated Visual Servoing Control for Mobile Robotic Manipulators without Joint-State Feedback

  • 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 disposing hazardous goods, mobile robots mayn't feed back the joint state of its equipped manipulator due to the damage of measuring devices. To avoid the disposition time delay caused by recalling the robot, an HRRC (human-robot-robot-cooperation) based uncalibrated visual servoing control system is presented for the mobile robotic manipulator without joint-state feedback. Firstly, the virtual exoskeleton (virtual model) for reflecting the joint state of the manipulator is set up through selecting the joint regions artificially with human-computer-interaction (HCI) input devices (such as mouse) on the monitoring picture captured by the camera from another mobile robot. Then, the virtual exoskeleton is combined with the multi-joint tracking algorithm, to steer joints of the manipulator and maintain the posture of the end-effector. In order to guide the virtual exoskeleton using artificial guidance points, the relationship between the terminal of the virtual exoskeleton and joint angles is mapped by GRNN (general regression neural network). In the peg-in-hole experiment, the end-effector completes the task under the artificial guidance, and the posture of the end-effector can be maintained within an error of ±1° using the proposed method. Comparatively, the end-effector can neither complete the task, nor maintain its posture using the conventional single joint control method. Results of the contrast experiment verify that the proposed control system can assist operators to intuitively use the manipulator to dispose targets without any feedback of the joint states, and to maintain the end-effector in a desired posture during the disposition.

     

/

返回文章
返回