Research On Multi-UAV Cooperative Tracking Ground Moving Target System

In recent years,with the continuous advancement of science and technology and the continuous growth of social demand,the small multi-rotor UAV industry has been able to flourish.Among many topics related to UAVs,as a sub-task in many UAV applications,target tracking has been widely concerned by scholars at home and abroad.In the course of performing a task,a gimbal camera mounted on a UAV is usually used to collect environmental images,then identify and track the target from them.However,when facing a non-cooperative target,the target may be temporarily lost from the camera's field of view,and the tracking task often fails because of this.In order to solve this problem,this article uses multiple UAVs to coordinately execute task to reduce the probability of target loss and ensure task continues.The specific research content includes the following aspects:1)First,the system functional architecture is designed according to the task requirements.The overall task is divided into three sub-tasks: single-UAV ground moving target positioning,multi-UAV target state fusion,and multi-UAV collaborative control.Then,in order to facilitate research and system verification,a software and hardware experimental platform was built based on the DJI Matrice 100 drone.2)Using a vision-based method,a single UAV is used to initially locate the ground moving target.First of all,considering the real-time and implementability of the system,the TLD algorithm is adopted as the visual target tracking algorithm,and the target feature information sharing strategy between the UAVs is added to realize the automatic initialization of the tracking task.Then according to the pixel coordinates of the target,combined with the state data of the UAV,coordinate calculation is performed to obtain the target position coordinates.Finally,the actual outdoor data is collected for experiments,and the experimental results prove the feasibility of the positioning system.3)Aiming at the problem that the target positioning results of multiple UAVs contain measurement errors and noise interferences,this article adopts distributed cubature information filter based on consensus theory to estimate target state.Firstly,a simulation experiment is designed in Matlab to verify the effectiveness of the algorithm,and a strategy is proposed to reduce the impact of network transmission packet loss on distributed computing.Then use C++ language to implement the algorithm,and compile it as a node under ROS,and conduct experiments to verify whether it is effective.4)Aiming at the problem of multi-UAV collaborative control under target tracking task,this paper proposes an upper-level control strategy including trajectory planning and tracking.At the level of trajectory planning,a polynomial function is used to describe the expected segmented trajectory of each UAV,and the polynomial coefficients are optimized taking into account the task objectives and actual constraints.At the trajectory tracking level,the expected trajectory is considered as a time-varying formation,and a distributed time-varying formation control algorithm based on consensus theory is used to track the expected trajectory of each UAV.The effectiveness of the proposed strategy is proved by Matlab/Simlink simulation experiments.Then,a distributed trajectory tracking control node is designed based on the established experimental platform,and a hardware-in-the-loop simulation experiment is performed to verify the trajectory tracking effect.