基于视觉信息的巡检机器人自主定位与路径规划研究

The Research on Autonomous Positioning and Path Planning of Inspection Robots Based on Visual Information

为了应对火电厂及热力泵站恶劣的巡检环境,该文提出一种基于视觉信息的巡检机器人控制系统。该系统以视觉同步定位与地图构建(visual simultaneous localization and mapping,vSLAM)技术为基础,融合点线特征信息进行巡检机器人的自主定位与地图构建,在构建成功的巡检环境模型中,使用改进的快速搜索随机树(rapidly-exploring random trees,RRT)算法完成巡检路径的自主规划。使用慕尼黑工业大学(technical university of Munich,TUM)数据集与ORB_SLAM2、PL-SLAM等系统进行对比,结果证明本文改进的融合点线特征的机器人定位与建图系统均衡地提高了定位精度与效率。使用仿真环境,将本文改进的巡检机器人路径规划算法与多种算法对比,结果证明本文的改进算法有效降低了规划路径长度、缩短了规划时间。

In order to cope with the harsh inspection environment of thermal power plants and thermal pumping stations,a kind of inspection robot control system based on visual information is proposed in this paper.Based on the technology of visual simultaneous localization and mapping(vSLAM),the system integrates the point and line feature information to carry out autonomous positioning and map construction of inspection robot.In the successfully constructed inspection environment model,the improved RRT algorithm is used to complete the autonomous planning of inspection path.Technical university of munich(TUM)data sets are used to compare the proposed positioning and mapping system with ORB_SLAM2 and PL-SLAM,and the results show that the improved robot positioning and mapping system fusing point and line features,which is proposed in this paper,improves the positioning accuracy and efficiency in a balanced way.The improved path planning algorithm proposed in this paper is compared with a variety of algorithms under simulated environments,and the results show that the improved algorithm effectively reduces the length of the planned path and shortens the planning time.