Estimation Of Relative Motion Parameters Based On Visual Feature Information Of Cooperative Targets

In order to meet the needs of the autonomous landing problem,this paper studies the relative motion parameter estimation algorithm based on the visual feature information of the cooperative target,which provides theoretical and technical support for the state estimation of the autonomous landing.The main research contents are summarized as follows:(1)The algorithms of cooperative target detection and recognition and visual feature extraction are studied.Firstly,according to the configuration design criteria of cooperative target,a new cooperative target composed of a circle and a T-shaped mark is designed,which is easy to identify and has sufficient visual feature information,and is used as an auxiliary carrier for navigation and positioning.Secondly,for the cooperative target acquisition image,the image preprocessing,ellipse target detection based on arc segment,cooperative target recognition based on Hu moment and other algorithms are studied Secondly,we study Harris adaptive threshold and corner detection and extraction algorithm based on convex hull analysis.Finally,we verify and analyze the algorithm by collecting images of cooperative targets under different conditions.The results show that the algorithm has high recognition rate and feature extraction accuracy.(2)The algorithm of relative pose estimation between carrier and cooperative target is studied.Firstly,the mathematical model and basic principle of relative pose estimation algorithm based on corner feature are analyzed,and the commonly used PNP algorithm is used to compare and analyze the corner feature of type mark.Secondly,the relative pose estimation algorithm based on circle and corner feature is derived,and the relative position and unit method based on single circle solution is eliminated by using the center projection coordinate The ambiguity of the vector solution,the course parameters are determined by any corner,all the navigation information is obtained,and the simulation experiment is used to verify and compare;finally,the actual data of the whole algorithm is verified and analyzed by the UAV landing experiment.The results show that the algorithm in this paper has relatively high navigation and positioning accuracy,and can meet the requirements of vehicle vision navigation autonomous landing.