Multi-target UAV Task Allocation And Tracking Syste

Multi-target tracking missions are common in modern warfare,such as aerial pursuits of the enemy or tracking strikes.However,the difficulty in acquiring target locations,the low efficiency of task allocation and the instability of tracking challenge the execution of tasks.Therefore,the combination of multiple UAVs with vision technology to track multiple targets over a large area has become a promising research direction.This paper takes the battlefield environment to study UAV tracking strategies.Three points are involved in the research:First,in solving the task allocation problem,a task model for multiple UAV target tracking is established and the selection criteria for targets are designed.This paper introduces the pigeon flock distribution algorithm and proposes a multi-level operator which takes advantage of information transfer at multiple levels to ensure a diversity and accuracy,so that a UAV can obtain the optimal initial target from a global perspective,to counteract its disadvantages,such as falling into local optimality easily.Since reassignment of local targets is difficult during the tracking process,this paper presents a parallel bidding contract network method,which uses a many-to-many assignment method under communication constraints to solve the problem of easy blocking.The simulation proves that the gain is increasing.Secondly,to overcome the loss problem caused by factors such as environmental changes and attitude adjustments,a scale-adaptive tracker with confidence evaluation mechanism is studied.For the target offset problem,the average peak correlation energy is used as the confidence determination criterion,so that the model can be updated with high confidence;based on the calculation method of chunked scale,the variable scale template is constructed in combination with the attitude characteristics of UAV flight to solve the scale misfit problem.The image tracking is projected into 3D space by image-based servo control to achieve effective and stable tracking of the target and to verify its tracking reliability.Finally,due to the challenges posed by the inability of MATLAB simulation to recreate physical factors,this paper constructs a complex environment over a city on the gazebo physical simulation platform to test the UAV’s ability to continuously track a target in complex terrain;and uses six multi-rotor UAVs to build a physical prototype experimental platform to verify its tracking and tasking capabilities in a real environment.The results show that the UAV can autonomously select the optimal target for tracking through the pre-assignment and reassignment process.