Research On Autonomous Landing System Of Quadrotor UAV In Power Inspection

The quadrotor offers major advantages includes outstanding maneuverability,low-cost and flexible flight characteristics when used for aerial photography,surveillance,disaster relief and power inspection.In order to solve the autonomous landing problems caused by the complex ground conditions,limited landing position,inaccuracy GPS signal in the special environment of the power inspection,a visual algorithm is proposed to improve the accuracy of landing.This method includes a searching and a landing part,the visual data is collected by onboard camera and the pose estimation is used to compute the relative position.The main research contents of this text are as follows:Firstly,in order to realize autonomous landing on designed position and reduce theAcomplexity of image processing,the shape "H" is used to be target image.A white disc is added to make separate from the background more easily.On the basis of the traditional image preprocessing,in order to solve the interference of the complex environment on the feature extraction,the operation of deleting the small connected domain is used,and the pixel points' value outside the pattern outline are unified to highlight the pattern.In the process of feature point extraction,after comparison with the common corner detection algorithm,this paper uses an angle sharpness method which combination the contour extraction and corner detection to determine whether a point is the required feature point.Secondly,according to the mathematical model of the quadrotor UAV and the camera,the required coordinate system is established and the conversion relationship between them is obtained.The pose estimation model is established by the coordinate relationship of the spatial points in the world coordinate system,the camera coordinate system and the image coordinate system.The camera's internal parameter matrix is obtained by camera calibration,and the homography matrix between the two frames of images is obtained.The homography matrix is decomposed,and the corner information obtained by the feature point extraction is combined to estimate the pose information of the drone with respect to the six degrees of freedom of the landing sign.Finally,the control algorithm is designed for autonomous landing system.In this paper,double closed-loop PID is designed for both position and attitude.The simulation experiment is completed by using MATLAB.Giving different initial value and control rate parameters,the rise time of the attitude angle is within 0.5s,and the stable value can be achieved to meet the control requirements,and the effectiveness of the algorithm is verified.At the same time,combined with the environmental characteristics of power inspection,a real wind model is simulated and the wind compensation algorithm is designed.Finally,the actual flight data of the drone is compared with the pose estimation data,in the meantime,the maximum error of the position and attitude estimation values is 0.52 m and 6.22°,which can basically meet the landing requirements of the drone,and the feasibility of the pose estimation algorithm is proved.