Research On Multi-UAV Cooperative Path Planning Based On Evolutionary Computation

With the development of the aviation industry and automation technology,UAVs are playing a more and more extensive role in the current social production activities of human beings,and UAVs can be seen in all major fields.The development of unmanned aerial vehicle technology continues to expand the application field of unmanned aerial vehicles.The applications in the civilian field are mainly concentrated in disaster relief,film and television shooting and publicity work,environmental protection,agricultural applications,transportation,etc.;in modern local wars Among them,UAVs have demonstrated their irreplaceable advantages,assisting in the completion of battlefield situation reconnaissance,precision strikes,cluster strikes and other combat missions.In many of the above fields,UAVs provide people with a more efficient,cheap,and low casualty way to complete a variety of different types of tasks.UAV path planning refers to finding the optimal flight path from the starting point to the target point that meets the limitations of the flight environment and the maneuverability of the UAV itself in a specific flight environment.It is the key technology of the UAV mission planning system.It is an effective means to ensure that UAVs improve application efficiency and successfully complete missions.Most of the existing path planning algorithms are proposed for robots moving on a two-dimensional plane.Three-dimensional path planning applied to UAVs faces a series of technical problems.Traditional path planning algorithms also have problems such as low efficiency and poor robustness.For the constrained large-scale combinatorial optimization problem of multi-UAV task allocation,it is difficult for the mathematical programming algorithm to obtain an accurate solution in a relatively short time.In addition,in an obstacled environment,it is still a challenge to quickly generate a flyable path for the drone.The coordination of multiple UAVs can double the application efficiency of UAVs.However,there are still few researches on path planning for coordination of multiple UAVs.In response to the above problems and challenges,the main research work and contributions of this article are as follows:First,in view of the basic and key issues in UAV path planning,this paper proposes a general problem modeling method.For the complex space environment of UAV flight,the physical obstacles and non-physical obstacles are modeled separately.For the problem of task allocation,general analysis methods and modeling methods are proposed,and general models are proposed for single-machine task allocation and multi-machine task allocation.In addition,for the path planning problem in two-dimensional and threedimensional space,a modeling method and solution representation method suitable for evolutionary computing are proposed,which completes important preparations for subsequent path planning.Second,for the large-scale combinatorial optimization problem of multi-UAV task allocation,an effective task allocation and traversal method based on simulated annealing is proposed to balance the tasks between drones and obtain satisfactory time.Resolution.Specifically,by adding virtual base nodes to the original vehicle routing problem model,it is easier to generate solutions suitable for heuristic algorithms.In addition,this chapter proposes the concept of a universal distance matrix,which can convert time constraints into spatial constraints,simplifying the programming model.Finally,a simulated annealing algorithm based on exchange and judgment strategy is proposed to improve the efficiency of generating feasible domain solutions.A large number of experiments and comparative studies in different simulation scenarios show that the algorithm has higher efficiency and accuracy compared with accurate algorithms and meta-heuristic algorithms.The results also inspired us about the characteristics of population-based algorithms to solve combinatorial discrete optimization problems.Third,in response to the problem of real-time generation of flying paths for UAVs in obstacle scenarios,this paper proposes a hybrid differential symbiosis search algorithm that combines the mutation strategy of differential evolution with an improved symbiosis search strategy..The algorithm retains the local search ability of the symbiotic organism search algorithm,and at the same time has excellent global search ability.The proposed algorithm introduces the concept of traction function to improve efficiency.In addition,the algorithm uses a perturbation strategy to further enhance its robustness.A large number of simulation experiments and comparative studies conducted in two-dimensional and three-dimensional scenes show that this algorithm has obvious advantages compared with other population-based evolutionary algorithms.Fourth,for the path planning problem of multiple UAVs arriving at the same time in a three-dimensional environment with multiple types of obstacles,this paper proposes a multi-UAV cooperative path planning algorithm based on adaptive differential evolution.Generate a more diverse initial population through Logistic chaotic mapping and the basic principles of artificial immune algorithms,combine the mutation strategies and parameter adaptive control strategies of two different adaptive differential evolution algorithms,and use the traction solution to make the population move towards the global optimal solution rapidly,thereby improving the search efficiency of the algorithm.Experiments in multiple simulation scenarios show that compared with other algorithms of the same type,the proposed algorithm is more efficient in solving this problem,the total cost of the generated path is lower,and the coordination between multiple machines is better.