Research On UAV Autonomous Landing Algorithm Based On Vision-guided

Unmanned aerial vehicle(uav) has the advantages of small size, low cost, and can adapt to various features of the work environment, is widely used in various fields. With the growing requirements of military, civilian and commerce, the research on UAVs have significantly development. During the UAV's flight, the stage of landing is the most prone to failure, so safely landing is a key point of UAV applications. Traditional navigation has the weak points of low precision, vulnerable electronic jamming and other defects, so, the navigation based on machine vision get more attention. This article focuses on the autonomous landing algorithm based on machine vision, and has field experiments made by UAV verify the availability of the algorithm, Three-dimensional scene simulation algorithm verify the accuracy of the algorithm. And the non-destructive Kalman filter algorithm applied to improve the accuracy and stability of the system.Firstly, we study the UAV image processing algorithms base on the visual navigation to detect edges and finish the matching between the landmarks and template, Landmarks again until you find the collaboration tracking using optical flow, it is a good way to improve the efficiency and real-time of the algorithm.Secondly, the paper studies the visual navigation control UAVs pose estimation models. According to the conversion relationship between each perspective projection model and coordinate system to obtain UAV six components overdetermined equations, and singular value decomposition method for solving this overdetermined equations.Then, the paper has field experiments made by UAV verify the availability of the algorithm, and introduces the synthetic vision three-dimensional modeling of the general process, and design algorithms to verify the accuracy of the algorithm.Finally, the paper studies the application of non-destructive Kalman filter algorithm and simulation algorithm is designed to verify that the Kalman filter algorithm is applied to visual navigation algorithms can make the pose estimation system more accurate and stable.