Research On The Technology Of Vertical Take-off And Landing Fixed-wing UAV Autonomous Landing Guidance Based On Vision

The Vertical Take-off and Landing Fixed-Wing UAV has the advantages of both high-altitude high-speed flight of conventional fixed wing drones and vertical takeoff and landing of multi rotor drones.At present,the electrically driven Vertical Take-off and Landing Fixed-Wing UAV is restricted by the Density dependence of the battery during flight.The wireless charging of the electrically driven Vertical Take-off and Landing Fixed-Wing UAV can properly make up for the limitation of small and light batteries,and the precise guidance of the Vertical Take-off and Landing Fixed-Wing UAV to the wireless charging platform can maximize the charging efficiency.Therefore,this project is based on visual measurement methods to study the autonomous landing guidance scheme of Vertical Take-off and Landing Fixed-Wing UAV and conduct experimental analysis on the entire system in simulation and physical environments.This has great engineering significance for achieving wireless charging of Vertical Take-off and Landing Fixed-Wing UAV to increase flight time.Firstly,the related coordinate system and the rotation matrix between the two coordinate systems involved in the visual guidance system are defined,and the parameters of the airborne camera are obtained through the three-dimensional to two-dimensional projection relationship.Nested cooperative targets that can estimate the relative position and attitude of multi altitude and multi angle are designed,and a vision based autonomous landing guidance scheme is proposed according to the working mode of the compound wing UAV.Secondly,to solve the problem of small target detection and occlusion in highaltitude environments,an ellipse detection algorithm based on arc clustering was designed.Obtain connected boundaries through image preprocessing and propose a series of constraint conditions to cluster arc segments.Use geometric and mathematical theorems to remove false parameters and fit real parameters.Compared with other detection algorithms,the detection accuracy is improved,the detection time is reduced,and the anti-occlusion performance is achieved.Subsequently,a continuous adaptive mean shift algorithm incorporating Kalman filtering was designed to address the difficulty of relocating targets after loss.The similarity evaluation coefficient is used to iteratively locate the target area through color features,and shape features are locally detected to correct the target area.At the same time,the search range is adaptively expanded.The occlusion coefficient is proposed to determine whether to use the Kalman filter prediction value as the optimal estimation value,and the guidance trajectory azimuth angle is calculated based on this optimal value.This algorithm can locate the target again after it is completely occluded.Furthermore,a segmented landing strategy was proposed.The pose parameters are calculated through the quadratic form mapping relationship of the space circle,and the ambiguity of the pose parameters is eliminated by using the straight line feature inside the circle.The relative yaw angle is calculated according to its asymmetric feature,and the landing rate when the cooperative feature cannot be obtained at the last stage of landing is designed.The pose solving algorithm is in the height range of 0.5m～1.5m,the position error is within 15 mm,and the attitude error is within 1°.Finally,establish a simulation and physical experimental environment to conduct engineering experiments on the visual landing guidance system.The feasibility of visual guidance scheme on stationary platform,uniform translational motion platform and deck motion platform is verified,and the detection performance of stationary and low-speed moving cooperative targets in high altitude in physical environment.At a flight altitude of 50 m,the lateral position estimation error of the parallel approach stage converged to about 1.2m,and the altitude error of the vertical landing stage converged to about 0.3m.The experimental results indicate that the visual guidance scheme designed in this paper is practical and reliable.