Research On Multi-UAV Collaborative Search Method Based On Local Information In Uncertain Environment

Unmanned Aerial Vehicle(UAV)technology has been rapidly developed in recent years.In the military and civilian fields,more and more UAV technologies are increasingly used in collaborative-target-search tasks,such as multi-UAV target search,exploration and rescue of casualties after disasters,transportation of goods and more.In the multi-UAV collaborative target search process,path planning has become a very important part,for finding targets and avoiding obstacles more effectively,and timely.These are the contents we need to study in the path planning.UAVs may be subject to various obstacles or interference of the no-fly zone,so that UAVs are unable to search as many targets as possible or the UAV’s coverage of the map is too small.These do not reach the mission indicators for the map search.In recent years,evolutionary algorithms have been widely used in path planning for multiple UAVs,mainly including: Genetic Algorithm(GA),Neural Network Algorithm(NEA),Immune Clonal Selection Algorithm(ICSA),Particle Swarm Optimization(PSO),and Ant Colony Optimization(ACO)and so on.These optimization algorithms are based on the task requirements of multiple UAVs to derive the corresponding system transfer function.The system transfer function is used to obtain the multi-UAVs task assignment and coordination framework.According to the characteristics of each algorithm,we optimize the overall mission planning of multiple-UAV.In this paper,we analyze genetic algorithm,Model Predictive Control(MPC)and greedy algorithm,and combine the three algorithms and add a new flight mechanism to it.The proposed new algorithm is named as an optimal dynamic compensation cooperative search algorithm based on region division(referred to as ODCA).In this paper,UAV’s flying environment of arbitrary size is modeled and simulated,and random small obstacles and large fixed obstacles are added.For complex environment,we consider not only the effectiveness of algorithm,but also the most reasonable paths of UAVs.When using ODCA for path planning and target search for multiple UAVs,the predictive control theory is used to predict the path of UAV firstly,and then the most suitable UAV flight path is obtained through the ODCA selection of suitable population.In the whole process,UAV is coordinated to inform each other about their own information,and through the cooperation,we can get a better result comparing with traditional algorithms.This paper summarizes the genetic algorithm,predictive control theory,and greedy algorithm firstly.Then the UAV flight environment,its own state,and search probability map are modeled.After modeling,we use traditional algorithm to carry out UAV path planning,and find out the problems of each algorithm in the simulation experiment diagram of path planning.In this paper,a single UAV path planning is carried out by each algorithm.The problem is found in a single UAV path planning and the multi-UAV search target path planning is made to make up for the existence of single UAV path planning.In the process of multi-UAVs collaborative search path planning,the problems appearing in each algorithm are found.The addition of flight mechanism in the traditional genetic algorithm makes the UAV get the maximum coverage in the path planning,so as to search for as many targets as possible.Under the algorithm after adding the new mechanism,the superiority of the new algorithm is obtained by comparing the experimental data of other algorithms by changing the relevant parameters.Finally,the research results are summarized and the outlook is made.Through the experimental results of this paper,it can be proved that ODCA can obtain better experimental performance than greedy algorithm,random algorithm and genetic algorithm.Simulation results verify the effectiveness of ODCA.