Visual-based Autonomous Navigation And Path Following For Inspection UAV

In recent years,intelligent robots have attracted more and more attention in the field of inspection of large facilities.Compared with manual inspection,intelligent robots have significant advantages in improving detection efficiency and reducing human costs and risks.Unmanned aerial vehicle(UAV)has been applied in power inspection,road and bridge monitoring and other fields because of its three-dimensional motion capability.However,the application of inspection UAV relies heavily on the navigation and positioning information provided by satellite.In the environment where satellite signals are weak or rejected,it is difficult to carry out inspections with expected effect.In addition,the normal path planning of inspection UAV usually adopts the form of pre-set waypoints in geographical coordinate system.In this way,even if the UAV cannot complete the effective detection of the target on the planned path,the UAV will not take remedial action.The effect of such inspection depends entirely on the quality of the prior path planning.In view of the above two shortcomings in the patrol UAV system,this paper studies the autonomous navigation and positioning,path recognition and following methods of UAV based on visual sensor,which improves the applicability and intelligence of the inspection UAV in various situations.Firstly,the application status of inspection UAV is analyzed in this paper.Combined with the navigation and positioning requirements of several practical application scenarios,a general optimization scheme of inspection UAV in indoor and outdoor environment is designed.The sensing principle and error characteristics of the main sensors in the scheme are analyzed,which provides a theoretical basis for the subsequent implementation of multi-sensor fusion algorithm.Secondly,this paper studies the detection and recognition methods of ground signs and relative positioning methods in UAV inspection system.On the one hand,combining the path characteristics and path model,a structured path detection method based on pixel distribution histogram is proposed,which can identify the structured path online in real time and calculate the relative offset between UAV and the path to be followed.The recognition algorithm of cooperative QR code is studied,and the relative position of UAV is calculated based on cooperative QR code.On this basis,an optimization algorithm of relative position estimation based on real-time attitude is proposed,and the effectiveness of this method is verified by experiments,which provides effective navigation information for autonomous path following of subsequent inspection UAV.Subsequently,this paper studies the navigation and positioning method of inspection UAV in the environment of satellite rejection.Firstly,the velocity estimation method based on optical flow is analyzed,and the traditional LK pyramid optical flow algorithm is improved and verified by experiments.Secondly,the inertia/optical flow integrated navigation algorithm is designed to further improve the velocity estimation accuracy of the UAV.The global position information of cooperative QR code estimation is introduced into the filter to improve the accuracy of measurement of UAV position and velocity.Finally,on the basis of path recognition and navigation positioning,this paper studies the vision-based UAV path following method.Aiming at the inspection task,the control strategy of UAV is studied and the control method of multi-stage PID series is designed.At the same time,based on the above research,the prototype of the inspection UAV is constructed,and the outdoor and indoor UAV inspection experiments are carried out,and the algorithm proposed in this paper is systematically validated.