一种输电线巡检机器人的自动抓线视觉伺服控制  

Visual Servo Control for Automatic Line-Grasping of a Power Transmission Line Inspection Robot

作　　者：郭伟斌[1,2] 王洪光[1] 姜勇[1] 孙鹏[1] 

机构地区：[1]中国科学院沈阳自动化研究所机器人学国家重点实验室,辽宁沈阳110016 [2]中国科学院研究生院,北京100049

出　　处：《机器人》2012年第5期620-627,共8页Robot

基　　金：国家863计划资助项目(2006AA04Z203);国家自然科学基金资助项目(60875082)

摘　　要：为了保证轮臂复合式巡检机器人自动越障时行走轮能够可靠抓线,提出一种基于图像的视觉伺服抓线控制方法.在图像空间定义输电线的偏距和偏角来表征图像特征的变化,并设计出位姿解耦伺服控制律.由于视觉雅可比矩阵与机器人状态有关且可能发生奇异,采用模糊方法进行机器人控制.通过实验确定模糊集合的论域,设计了两层模糊控制器并编制了自动抓线程序.进行了上百次实验,其抓线控制的偏距和偏角误差分别在25像素和2以内,能够可靠抓线,实际越障中只需小于40s的时间即可完成抓线.结果表明该方法具有准确、可靠、效率高的特点,能够满足准确、快速、可靠抓线的任务要求.To guarantee that the wheel can grasp the power line reliably when the wheel-arm inspection robot crosses obstacles automatically, an image based visual servo control algorithm for line-grasping is presented. The offset distance and declination of power line, defined in image space, are used for describing the changes of image features. Then, the position and orientation decoupling control algorithm is established. For the visual Jacobian is decided by the robot status and may be singular, therefore a fuzzy method is applied to robot control. The universes of fuzzy sets are specified by experiments. Then a two-layer fuzzy controller is formulated and an automatic line-grasping program is developed. In more than 100 experiments, the offset distance and declination errors of line grasping are less than 25 pixels and 2~, reliable line-grasping is realized, and the robot can grasp the power line in 40 s when crossing obstacles. The results show that the image based visual servo control algorithm is precise, reliable and efficient, and can satisfy the requirements of the accurate, high-speed and stable line-grasping tasks.

关 键 词：视觉伺服 自动越障 巡检机器人 架空输电线路 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置] TM7[自动化与计算机技术—控制科学与工程]