Research On Key Technology Of Aircraft Landing Based On Vision-assisted Integrated Navigation

With the development of science and social productivity,the demands for autonomous landing on aircraft increasingly improve.The navigation technology establishes the space geometry relation between the aircraft and environment,realizing the landing guidance for aircraft,which is an important support for the development of intelligent and unmanned flight.Facing to the process of autonomous landing of fixed wing UAV,based on the existing GNSS/INS integrated navigation,this dissertation focus on the vision measurement of aircraft's position and attitude based on airborne and landing field,and its combination with GNSS/INS integrated navigation,in the process of which,the key problems of detection and extraction of mark image,error analysis and others are researched.Through the establishment of the airborne/landing field joint vision-assisted integrated navigation system,the safe and reliable autonomous landing can be realized,aiming to provide support for the development of autonomous landing of the related UAV.The main work and innovation point are as follows:(1)The real-time monitoring method for the pose of the multi camera in the outdoor environment is studied.A two-step combined online monitoring of the position and attitude of the multi camera is proposed.Firstly,the initial value is obtained through the collinear intersection equation in quaternion form,then,which is refined through sparse beam adjustment based on LM algorithm.The practical and simulation experiment indicate that the algorithm can realize the online monitoring of the pose of multi camera,and has strong robustness to the system noise,which ensure the stability of multi camera position and attitude monitoring.(2)The method of feature detection and extraction in dynamic flight environment is studied.The GNSS/INS-assisted active mark center fast extraction algorithm is proposed.Firstly,the position solution of GNSS/INS is used as the coordinates of the mark point,and then realize the measurement of initial image point coordinate based on collinear equation,Secondly,based on the feature distribution of the mark,the influence of smear is eliminated using the local background subtraction and the contour segmentation algorithm,and then realizing the sub-pixel extraction of non-standard ellipse centers,which ensuring the accuracy and real-time positioning.(3)The vision-assisted GNSS/INS integrated navigation algorithm based on geometric constraint is studied.Based on the analysis of the error characteristics of vision navigation and positioning,the joint strategy of visual assistant system is designed,and then,an aircraft and landing field joint vision-assisted GNSS/INS integrated navigation algorithm is proposed,taking into account the visual radial observation is not sensitive,as well as the inconsistency between the data sampling,a multi-scale EKF integrated navigation filtering model based on visual radial constraint is proposed.The integrated navigation experiments are carried out under three different conditions,The results show that the proposed algorithm is more accurate than the GNSS/INS integrated navigation and the airborne vision-assisted GNSS/INS integrated navigation,avoiding the influence of the lack of measurement.