Research On The Autonomous Precision Landing System Of UAV Based On Visual Navigation

In recent years,with the rapid development of UAVs and related technologies,the UAV have been used in all walks of life.Improving the autonomous operation capability of has always been the focus of attention.As the most complex landing process in autonomous flight missions,it also faces new challenges.Therefore,the research on the autonomous landing technology of quadrotor UAV is particularly important.The autonomous landing of the UAV not only requires stable control of its own attitude,but also needs to complete high-precision positioning.Aiming at the problems of low accuracy and excessive error of current traditional navigation methods such as GPS,based on the existing research of the research group and combined with visual navigation technology,this paper designs a set of autonomous landing system based on ArUco logo,and conducts experimental verification.The main research contents of this paper are as follows:(1)Research and propose a cooperative nested landing sign based on ArUco QR code.Relevant optimization has been carried out on the original detection algorithm.The logo recognition is completed through seven steps of image grayscale,thresholding,contour extraction,quadrilateral frame selection,perspective change,corner extraction,and id information extraction.The problem of unsteady recognition improves the accuracy of landing sign recognition.(2)Research and improve the pose solving algorithm.By fusing corner point information and PnP positioning algorithm,the relative attitude estimation value between the current UAV and the landing marker is obtained by solving the feature point coordinates.Aiming at the partial occlusion of the landing sign,a vertex completion algorithm based on PnP is proposed to improve the anti-interference ability of the UAV.Finally,an error model based on linear regression is proposed to further compensate the positioning information and provide accurate position information for the autonomous landing of the UAV.(3)Research and propose a vertical landing strategy.Through vertical landing,the interference of ground effect to the UAV is reduced,and the landing efficiency of dynamic target tracking is improved.Finally,the simulation experiment was completed by building a simulation platform based on Gazebo,and the real flight experiment was carried out at the same time.The experimental results show that the UAV autonomous landing system proposed in this paper ensures the stability of the UAV flight and effectively improves the landing accuracy.The average detection time of signs is stable between 30 and 40ms,and the average landing accuracy is controlled within 5cm.