Research On Multi-UAV Trajectory Planning And Coordination Algorithm Based On Rolling Optimization

UAVs(unmanned aerial vehicles)have the advantages of high mobility and low cost,especially no pilot driving,making it widely used in military,industrial and consumer fields.With the improvement of flight control and computer technology,it is possible for UAVs to perform tasks autonomously and cooperatively.By improving the autonomy of UAVs,we can reduce labor cost and lower the dependence on communication quality.And the multi-UAV cooperation can improve the performance of UAVs.Therefore,the realization of UAV autonomous cooperative technology can greatly improve the application ability of UAV.However,the autonomous coordination ability of UAVs is weak at present.In order to improve the autonomous abilit y of UAV,a complete path planning scheme is needed.So this dissertation studies single UAV path planning methods,and based on that,we study the optimization method of collision avoidance for multiple UAVs,and form a complete set of multiple UAV path planning and coordination schemes.The main contents are as follows:Due to the fact that the real environment is often known as rough environmental information,the fine environmental information is unknown and the environment changes dynamically,it`s difficult for traditional methods to plan path in such an environment.In view of the defects of traditional methods,this dissertation adopts double rolling windows scheme to integrate the global and local path planning methods,which makes full use of global and local information while ensuring the optimality and adaptabili ty.Built on the theoretical and experimental analysis,it is proved that the PRM(Probabilistic Roadmap Method)+ Dijkstra algorithms are more suitable for the UAV global path planning than the grids method + A* algorithms.So in this dissertation,the PRM method is used to model the map,the Dijkstra algorithm is used to generate the global trajectory,and among the global path points,a Rapidly-Exploring Random Trees(RRT)method is used to avoid unknown obstacles and smooth path.The simulation results demonstrate that the fusion algorithm scheme is feasible and effective.Aiming at the special needs of UAV local path planning,the RRT algorithm is optimized by exploring with variable length and variable probability,and the simulation results show that this scheme is more efficient than the basic RRT algorithm.This dissertation models the collision avoidance problem of multi-UAV operated in the same airspace.A distributed coordination algorithm based on predictive state is adopted to solve the problem.Based on the predetermined path planing by above path planning scheme,this algorithm is applied to optimize the predetermined path of all UAVs to avoid collision,which achieves the combination of path planning and coordination.This dissertation uses a distributed method to solve the problem.Compared with the centralized scheme,the distributed scheme has the advantages of reducing the complexity of problem and adapting the changing network structure of UAVs.After modeling,it is found that multi-UAV collision avoidance problem is a multi-objective and multi-constraint problem.This dissertation uses PSO(Particle Swarm Optimization)algorithm which can efficiently search solution to solve this problem.The simulation results show that the proposed scheme can guarantee the safe flight of UAVs when the forecasting step length is sufficient.