变电站设备巡检机器人视觉导航方法

Visual Navigation Method of Substation Patrol Robot

针对变电站巡检工作的实际需求,介绍了一种基于轨线引导的巡检机器人单目视觉导航方法。对于预先设置的引导轨线和停靠位标识,采用OpenCV和DirectShow软件库来开发视频采集程序,实现道路视频信息的实时采集;通过HSI模型对图像进行颜色提取,再采用数学形态学处理去除干扰信息;最终采用统计8-连通区的方法将导航线和停靠位标识分割提取出来,用采样估算方法计算导航线参数,通过比例系数融合导航参数以控制两驱动轮电机实现机器人的运动控制。实验表明,该方法实现了机器人的自主导航,能够适应绝大部分光照强度的环境,并具有较好的实时性和可靠性。

In view of actual demand of circular inspection of substation equipments, a visual navigation method based on trajectory guidance for monocular patrol robot is presented. For preinstalled guide trajectory and stop marks, the video capture program is developed by OpenCV and DirectShow software library to implement real-time acquisition of video information of road; the target colour is extracted by HSI model, then the interference information is removed by mathematical morphology; finally the trajectory and stop marks are segmented and extracted by calculating the 8-adjacent connected areas, and the trajectory parameters are calculated by sampling-estimation method, the electric motor for two driving wheel is controlled by navigational parameters which are fused by proportional coefficient to implement the motion control of patrol robot. Experimental results show that the presented method possesses better real-time performance and reliability and the autonomous navigation of patrol robot can be realized by this method.