Research On The Method Of Multi-UAV Cooperative Target Search And Tracking

With the widespread application of UAVs(Unmanned Aerial Vehicles)in the military and civilian fields,as an important part of UAV tasks,target search and tracking have become one of the hot spots in unmanned systems and cooperative technology research.Existing research does not have a thorough understanding of the complexity of the task environment,the uncertainty of target information,and the unreliability of target detection algorithms.Their research results are still far from practical applications.Facing actual needs,starting with the existing experimental conditions,this paper has carried out the research work of target search and tracking as follows:(1)Facing the path planning problem of multi UAV target search in obstacle environment,aiming at the complexity of task environment and the limited perception and movement ability of UAV,this paper designs single UAV performance index and multiUAV cooperative performance index based on target probability map(TPM)and digital pheromone map(TPM).Based on this,a cooperative decision search method(CAS)is designed by using rolling horizon control strategy.The CAS method solves the local optimal control input based on the single UAV performance index,obtains the multi-UAV cooperative performance index and solves the global optimal control input.Through rolling horizon control and deepening iterative optimization strategy,the CAS method ensures the optimality of the final multi-UAV control input.(2)For the target tracking task based on monocular vision localization,considering the unreliability of target detection algorithm,this paper designs the error evaluation function based on monocular vision localization,and proposes a multi-UAV cooperative localization method.Through three types of reasonableness testing methods: time,displacement and speed,the method filters out abnormal data generated by the mis detection of the target detection algorithm,and fits the current location of the target according to the historical trajectory when there is no reasonable data.Based on the value of the evaluation function,the method filters and sorts the monocular visual localization results of multiple UAVs in a distributed decision environment,and realizes that multiple UAVs can autonomously generate a consistent localization result.(3)According to the different requirements of target search and tracking on mission environment,this paper constructs distributed multi-UAV cooperative search simulation system and cooperative target tracking prototype system respectively.The simulation system simulates the distributed decision environment of multiple UAVs when performing search tasks,and verifies the performance of the CAS method.By giving grid maps with different pheromone concentrations and target probabilities,multiple UAVs can spontaneously plan search paths.Based on monocular vision localization method,the prototype system uses onboard computer,high-precision camera and locator to realize the position perception,communication,target detection and localization of multiple UAVs required for target tracking task,and verifies the superiority,stability and expansion of cooperative localization method.