基于一致性交叉的带电作业机器人目标定位  

Target Localization of Live Working Robot Based on Consensus Crossover

作　　者：张富荣 番禹 沈忠亮 陈兴国 ZHANG Fu-rong;FAN Yu;SHEN Zhong-liang;CHEN Xing-guo(Dehong Power Supply Bureau,Yunnan Power Grid Co.Ltd.,Dehong 678400,China)

机构地区：[1]云南电网有限责任公司德宏供电局,云南德宏傣族景颇族自治州芒市678400

出　　处：《计算机仿真》2025年第2期472-476,508,共6页Computer Simulation

基　　金：中国南方电网有限责任公司科技项目,项目名称:220kV及以下电压等级带电作业便携式平台技术研究(YNKJXM20220063)。

摘　　要：在复杂带电作业环境中,可能存在多个相似或混淆的目标,导致在机器人视觉下难以辨别像素间的相似度,易造成误识别或混淆识别,影响目标定位精度。针对这一问题,提出基于Census算法的带电作业机器人目标定位方法。结合像素点间RGB颜色绝对差值与Census变换值,计算作业目标图像与历史记录图像像素之间的相似度,帮助机器人确定目标在当前环境中的位置和方向。利用八方向半全局视差技术获取目标的初始稠密视差图,通过左右一致性交叉检测得到初始视图中不稳定的视差值,再运用视差优化算法优化不稳定视差,确定候选区域;采用二次成像法融合目标测量值和超声波测量值,恢复目标直径信息。在目标尺寸已知的情况下,通过三维定位方法定位目标的真实位置信息。仿真结果表明:该方法可以获取精准的带电作业机器人目标定位结果,可以为机器人带电作业提供有效保障。In complex live working environments,there may be many similar or confused targets,which makes it difficult to distinguish the similarity between pixels under robot vision,leading to misidentification or confusion,and affecting the accuracy of target localization.Aiming at this problem,a target location method of live working robot based on Census algorithm is proposed.Combining the absolute difference of RGB color between pixels and Census transform value,the similarity between the pixels of the job target image and the historical record image is calculated to help the robot determine the position and direction of the target in the current environment.The initial dense parallax map of the target is obtained by using the eight-direction semi-global parallax technology,and the unstable parallax value in the initial view is obtained by the cross detection of left and right consistency,and then the unstable parallax is optimized by using the parallax optimization algorithm to determine the candidate area.The second imaging method is used to fuse the measured value of the target and the measured value of the ultrasonic wave to recover the diameter information of the target.When the size of the target is known,the real position information of the target is located by three-dimensional positioning method.The simulation results show that this method can obtain accurate target positioning results of the robot with live working,and can provide effective guarantee for the robot with live working.

关 键 词：Census算法 带电作业机器人 目标定位 相似度 视差值 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]