On The UAV Localization And Capture Based On Binocular Vision Under Complex Background

With the rapid development of Motor drive,automatic control and sensor technologies,UAV system has been widely used in various fields such as military,agriculture,photography and so on.Multi-rotor UAV is capable of autonomously completing various aerial tasks because of its simple operation and fixed hover.After the completion of the mission,the multi-rotor UAV cannot land on any complex terrain,which is a very important link for its safety recovery.This paper proposes a 6-UPS parallel machine as the platform,and integrates the binocular vision positioning technology to carry out the research on the tracking and positioning of multi-rotor UAV in complex environment.The main research of this thesis are as follows:Firstly,the system of UAV capture robot is designed and developed based on PAC(Programmable Automation Controller,PAC)and STC15.Using binocular stereo vision as a measuring tool,the position and position of UAV are measured and positioned in real time,and the transformation relationship between the camera coordinate system and the mechanical grasp coordinate system is established.Zhang Zhengyou camera calibration method was used to complete the binocular system calibration,SVD decomposition method was used to derive the hand-eye calibration model,and finally completed the system calibration.Secondly,the image enhancement techniques in various complex weather environments are investigated,and an adaptive image enhancement method based on morphology is proposed for several scenes of foggy,sunny and rainy and snowy days.First,the dark primary color prior theory is used to complete the image defogging,and then the morphology-based adaptive image enhancement method is used for the image defogging,and the comparison with the common image enhancement methods is made.Finally,the algorithm in this paper is more in line with the current used scene.In order to make the local feature enhancement more obvious,the YOLOv4 model is employed to detect the local feature,and then the local feature is enhanced.Furthermore,a dynamic tracking technology based on the ArUco Marker and sparse optical flow is proposed.The ArUco Marker is taken as the feature point of sparse optical flow to track the target,and then the motion velocity of the target is calculated.In order to further verify the effectiveness of the algorithm,the UAV virtual simulation platform is created based on the Unity rendering engine.The entire platform can be created through parametric design,and the simulation movement of the UAV and parallel robot can be controlled through C#.Finally,through the simulation platform,the UAV dynamic tracking simulation experiment is completed.Finally,an experimental prototype of the UAV capture platform robot was fabricated,and the control software platform of the robot is developed.A series of capture experiments of the UAV were successfully carried out,and the effectiveness of the scheme and algorithm was verified.