Research On UAV Visual Navigation Tracking And Landing Method Under Similar Target Interference

The problem of autonomous tracking and landing of UAVs for cooperative targets has been a hot research topic in the field of UAVs.As the current missions are increasingly inclined to GPS-free or weak GPS environments,UAV tracking and landing tasks through visual navigation have become the development direction of UAV autonomous control.The visual navigation method often encounters the problem of similar target interference in the tracking and landing process,i.e.,there is a cooperative target to be tracked and multiple interfering targets with similar appearance in the field of view.With the rapid development of computer vision,deep learning-based target detection and tracking methods have greater advantages in detecting and tracking cooperative targets in complex scenes,and are suitable for UAV tracking and landing tasks.This paper addresses the problem of UAV tracking and landing of cooperative targets under similar target interference,and establishes a complete UAV tracking and landing control strategy based on vision navigation from hardware system construction to deep learning algorithm design,which realizes the autonomous tracking and landing of UAV under similar target interference and improves the stability of UAV intelligent flight.The main research contents of this paper are as follows:1.Firstly,the design of cooperative target is completed,and further based on the quadrotor UAV,a complete system solution is designed for the tracking and landing problem of cooperative target under similar target interference,including the identification and localization system of cooperative target,UAV tracking and landing system.Based on the performance requirements of the above system for vision sensors and airborne computers,the selection of hardware platform and the construction of overall hardware platform are completed.The nonlinear dynamic equations of the UAV were derived,which laid the theoretical foundation for the subsequent research of the relative position estimation algorithm and the design of the sliding mode controller.2.For the cooperative target tracking problem under similar target interference,the cooperative target tracking algorithm is proposed based on Yolov5 and SiamRPN algorithms by combining Yolov5 and SiamRPN algorithms through IOU,distance threshold and template matching.The effectiveness of the algorithm is verified by cooperative target tracking experiments in a simulation environment.3.For the control problem of UAV tracking and landing based on visual navigation,the on-board camera internal reference calibration was completed,and the algorithm of relative position estimation between the cooperative target and the UAV was developed to obtain the desired position of the UAV in the process of tracking and landing.Based on the sliding mode controller,the internal and external loop control system of the UAV was designed to realize the smooth and fast flight test of the UAV from the current position to the desired position.Finally,the simulation experimental environment of the overall system and the real experimental environment are built,and the effectiveness of the method is verified by simulation and real experiments.