基于单眼视觉的巡检无人机避障策略研究  

Research on Obstacle Avoidance Strategy of Patrol UAV Based on Monocular Vision

作　　者：曹凯 李宁 王立志 李波 赵娜娜 CAO Kai;LI Ning;WANG Lizhi;LI Bo;ZHAO Nana(Ultra High Voltage Company,State Grid Ningxia Electric Power Co.,Ltd.,Yinchuan Ningxia 750011,China)

机构地区：[1]国网宁夏电力有限公司超高压公司,宁夏银川750011

出　　处：《电子器件》2025年第1期86-90,共5页Chinese Journal of Electron Devices

基　　金：国网宁夏电力公司科技项目(5229CG22000M)。

摘　　要：现代无人机已经具备一定的自主避障能力,但在复杂的变电站内部环境中,诸如电缆、设备等障碍物密集的区域,无人机的避障性能仍然存在挑战。提出了一种重构障碍物三维信息的方法,使无人机(UAV)能够检测并避开其飞行路径上存在的障碍物。首先,使用多尺度定向斑块(MOPS)检测出障碍物的大致轮廓。然后,通过标度不变特征变换(SIFT)算法计算障碍物内部轮廓特征点的空间坐标。最后,合并MOPS结果和SIFT算法的结果以显示障碍物的三维信息。所提出的方法只重建了障碍物的近似轮廓,因此提高了计算速度。此外,由于轮廓信息是通过SIFT特征点合并的,可以得到与障碍物有关的详细的三维信息。Only relying on the UAV camera,it is difficult for the UAV to avoid unexpected obstacles on the preset route during patrol in-spection in the substation.A method to recognize the three-dimensional information of obstacles is proposed,so that UAV can detect and avoid obstacles in its flight path.First,the rough contour of the obstacle is detected by using multiscale oriented patches(MOPS).Then,scale invariant feature transformation(SIFT)algorithm is used to calculate the spatial coordinates of the feature points of the inter-nal contour of the obstacle.Finally,the results of MOPS and SIFT algorithm are combined to display the three-dimensional information of obstacles.The method proposed only reconstructs the approximate contour of the obstacle,thus improving the calculation speed.In addition,because the contour information is merged through SIFT feature points,detailed three-dimensional information related to obsta-cles can be obtained.

关 键 词：无人机避障 障碍重建 多尺度定向斑块 标度不变特征变换 

分 类 号：TN911.73[电子电信—通信与信息系统] TP242[电子电信—信息与通信工程]