Multi-view Target Tracking Algorithm For Multi-UAV Collaboration

In recent years,with the rapid development of related technologies of the Rotorcraft Unmanned Aerial Vehicle(hereinafter referred to as UAV),which are widely used in all fields,such as military and civilian fields in daily life,aerospace and military affairs.Among them,the use of UAV to track and locate the target of the ground has always been a hot research at home and abroad.By continuing to locate and track ground targets and providing information about the targets to the UAV,to ensure that the UAV to complete the follow-up related tasks.Taking into account the operational efficiency and target location accuracy,UAVs are beginning to evolve from multi-UAV co-operation to the stand-alone operation of single UAVs.In there,we firstly use the single UAV in the room to locate and track the ground targets in real time.Then,based on this,we apply a number of UAVs to this scene and study and realize from the visual positioning and the target tracking.The main research contents include:This paper presents a monocular visual indoor target location algorithm which is suitable for on-line application of UAV based on autonomous building of UAV platform and motion capture system.First of all,through the airborne camera,the real-time image information of the marker on the target car is obtained.Then,using the semi-direct vision odometer which is combined with the actual needs of this article to be modified,the position and pose of the camera and the position of the feature points on the marker are calculated respectively,while constantly optimizing.Finally,merge it with the UAV position calculated by the motion capture system in a custom world coordinate system.After a series of calculations,the coordinates of the center of the marker in the world coordinate system are obtained.Experiments show that the algorithm has the advantages of real-time and high precision of target location.In this paper,we propose a autonomous tracking algorithm of UAV to target,which is based on kernel correlation filtering tracking algorithm and image visual servoing.The algorithm mainly uses the change of the center point of the box which is select the target and the generated pixel deviation as the UAV speed control volume.After processing,it will be sent to the bottom flight control in the form of ROS topic to achieve the goal of tracking the target autonomously.However,given the uncertainties,we need to limit the amount of speed control to ensure flight safety.Experiments show that UAV tracking is stable and works well.In this paper,complementary fusion filtering and consistency control algorithm are used to propose a multi-UAV co-location and autonomous tracking algorithm.Mainly through the coordinated control of the linear velocity of the two UAVs in the X and Y directions.Ultimately,the UAV can track the target at the same speed,and then the positioning data will be automatically saved for later fusion.Experiments show that the multi-machine tracking effect is good and the data accuracy after fusion is higher.