Research On UAV Obstacle Avoidance Technology Based On Video Images

Unmanned Aerial Vehicle（UAV）are increasingly used in military,agricultural,logistics,surveying,and search and rescue missions.There are many uncertainties in the implementation of the mission of the UAV,when the UAV collides with the obstacle,the UAV bomb accident occurs.Therefore,effective methods need to be taken to make the UAV have the ability to autonomously avoid obstacles.In this paper,aiming at UAV avoid the obstacle,the obstacle information acquisition and UAV obstacle avoidance path planning methods are researched.The main contents of the study are as follows:（1）The overall scheme of UAV obstacle avoidance system is researched.Aiming at the problems faced by UAV obstacle avoidance,according to the actual needs of UAV obstacle avoidance,the obstacle detection methods,the advantages and disadvantages of obstacle information acquisition technology of various mainstream methods are studied,and the obstacle information acquisition method by the use of binocular vision is determined determines the use of binocular vision.The overall scheme of UAV obstacle avoidance is designed,which is mainly divided into three parts: binocular stereo visual positioning system,obstacle motion state estimation system,and UAV obstacle avoidance strategy.In addition,the UAV positioning and obstacle positioning methods are studied,and the conversion relationship of coordinate system the obstacle avoidance system and the three-dimensional position positioning formula of the obstacle based on the least squares method are derived,this is a foundation for the research of the UAV obstacle avoidance method.（2）The obstacle localization technology based on binocular stereo vision is researched.The camera imaging model and camera calibration method are elaborated,and the camera calibration experiment is completed on this basis.Considering the real-time requirements of UAV obstacle avoidance,the feature point extraction method combined with obstacle segmentation and ORB stereo matching algorithm is studied,and the accurate matching of feature points is realized by extracting obstacle feature information and combining the polar line constraint criterion to eliminate false matching points.According to the feature point information,the center position and dimension of the obstacle are obtained.Experimental results verify the feasibility of the method.（3）The estimation method of the motion state of obstacles based on Kalman filter is studied.In view of the unknown motion state of obstacles,the problem that all obstacles are regarded as static obstacles will cause the UAV to avoid obstacles failed,and the method of determining the dynamic static of obstacles and the estimation method of dynamic obstacle movement state are studied.According to the position vectors of obstacles at different times obtained by binocular stereo vision,a judgment criterion is established to determine the state of the obstacle.The Kalman filtering algorithm is used to estimate the motion state of dynamic obstacles,which lays a foundation for the planning of the obstacle avoidance path of the UAV.Experimental results show that the introduction of obstacle motion estimation significantly improves the accuracy of obstacle information acquisition.（4）The UAV obstacle avoidance path planning algorithm is researched.Aiming at the complex shape of obstacles and the uncertainty of motion,the algorithm of UAV obstacle avoidance path planning is studied.Through the binocular visual estimation of the obstacle dimensions,a three-dimensional expansion model of the corresponding obstacle is established,and the artificial potential field method is used to plan the obstacle avoidance path of the UAV.According to the local minima traps,unreachable targets,long obstacle avoidance paths and dynamic obstacle avoidance failures existing in the traditional obstacle avoidance algorithms,an improved algorithm for UAV obstacle avoidance path planning based on artificial potential field method is designed.Experimental results show that the improved artificial potential field method effectively solves the defects of the traditional algorithm and improves the reliability of obstacle avoidance.