Standoff Tracking And Control Method For Cooperative Uavs To Follow Ground Target

With the progress of science and technology,modern war is gradually moving from mechanization and informationization to intelligentization,and the form of war is developing toward unmanned and clustering.UAV has also gradually stepped onto the stage of war conflicts,and played an important role in the war between Azerbaijan and Armenia in 2020,playing a decisive role in the war.It is greatly theoretical significance and application value to carry out the research for UAVs to detect and track ground target.In this context,this paper studies on standoff tracking and control method for cooperative UAVs to follow ground target,breaks through the key technologies for UAV to detect ground target based on deep neural network and UAVs to cooperative position and track ground target,and realizes the cooperative UAVs standoff tracking ground target.Firstly,the deep learning-based UAV visual target detection and tracking method is studied.Based on the YOLO(You Only Look Once)deep neural network,the UAV ground target detection is realized.On the basis of the target detection,an integrated framework of target detection and tracking that integrates Yolov3 and Siam Mask is established to realize the detection and continuous tracking of the target in sequence images.Secondly,the method of UAVs cooperative target location based on distributed filtering is studied.A distributed control system based on state decision-making machine is designed to make the UAV change its state through different triggering events to accomplish the task of co-locating the target.The results of UAV target recognition are solved by image calculation as quantity measurement,and the motion of ground targets is modeled as the equation of state.The EKF(Extended Kalman Filter)positioning is implemented for each UAV respectively to obtain the target position estimation.An effective state vector fusion method is presented to optimize the location information from UAVs.The distributed filtering method can improve the redundancy of the system.Simulation experiments verify the effectiveness of the proposed co-location method.Thirdly,the research based on Lyapunov vector field for cooperative UAVs to track ground target is carried.Construct the navigation vector field for UAVs’ flight,for a single UAV circling the ground stationary target and cooperative UAVs tracking the ground moving target,the tracking controllers are designed respectively,on the basis of cooperative UAVs target positioning,an integrated simulation experiment is carried out for two UAVs tracking ground moving target.Finally,a visual simulation platform based on Air Sim is built,and the overall architecture of the simulation platform is designed.The terrain environment database and target model database are established based on UE4,and Air Sim is used as a quadrotor simulator in the virtual scene of the database.The target detection subsystem,target positioning subsystem and tracking control subsystem are designed respectively.The platform’s functions include automatic target recognition,hovering over a preset area and hovering over a stationary target on the ground.