A Study On Multi-targets Searching By Multiple UAVs

In the target search mission of drones,the limited load of single drones and limited detection accuracy limit the role of drones in target search tasks.Because the environment of different search tasks is very different,if the environmental information cannot be effectively utilized,it will seriously hinder the completion of the search task by the drone.Many current search strategies do not consider the mobility of the target and are difficult to apply to the actual search scenario.For the search task itself,there is a contradiction between the prec ise search target and the shortest time coverage search.Limit by the accuracy of the sensor,accurate search will inevitably be repeatedly confirmed,resulting in increased search time;short-time coverage search will not guarantee the accuracy of the sea rch.In order to solve these problems,this thesis aims at the contradiction between the accuracy and time of the precise search and the cover search in the search task,and studies the multi-UAV multi-moving target as the search target,and constructs the environmental information model to help the unmanned The machine understands the environment and designs a variety of drone search strategies to balance this contradiction.The environmental information model is the link between the drone and the environment,and is an essential foundation for the UAV search problem.With the development of UAV technology and the flexibility of the UAV,more and more search tasks can be implemented with UAVs.This paper addresses the contradiction between precision and time in the precise search and coverage search,with multiple moving targets as the search target,a variety of UAV search strategies were designed to balance this contradiction.For multi-UAV search problems,rationally model the environment.Search map is discretely rasterized,and the corresponding probability map update model and the deterministic update model are introduced to update the environment information.Aiming at the mobility of search targets,a hormone map model was designed to guide the UAV to return to the necessary grid units,and avoid multiple drones to search the same area,which improved the search efficiency.The mathematical model of the scene is established for the scene of accurate search.The environmental information model is used t o realize the interaction between the UAV and the environment.A variety of search correlation functions are designed to evaluate the search decision.The distributed model predictive control model is used to solve the decision information of the UAV syste m.A hybrid particle swarm optimization algorithm is designed to solve the model.The feasibility of the strategy is verified by simulation experiments.For this scenario,this strategy is compared with other search strategies,which verifies that the strategy can effectively improve search efficiency and achieve accurate search of mobile targets.The mathematical model for the scene is established for the scene covering the search.Starting from the basic shortest path problem,a centralized search strategy combining multi-traveler problem and balanced path length is designed to realize fast and comprehensive coverage of search maps by UAV clusters,and at the same time balance the search path of each UAV.Aiming at the disadvantages of high complexity and inability to dynamically allocate online,a distributed search strategy based on region allocation is designed.It mainly includes three parts: grid regionalization,regional allocation and regional path planning.Aiming at grid regionalization,a reverse K-nearest neighbor algorithm is designed to realize reasonable division of grid cells and replace grid cells with regions to reduce computational complexity.For the regional allocation,a model predictive control strategy is introduced to achieve dynamic online area allocation.The efficiency of the centralized search strategy in covering the search scenario is verified by simulation.It verifies the advantages of distributed strategy in algorithm complexity.The process of dynamic search indicates the rob ustness of the distributed strategy.