| In this paper,cooperative path planning problem of multi quadrotor aircraft in twodimensional plane is analyzed and the algorithm is designedFirstly,the coordinate system,the transformation relationship between coordinates and the kinematics and dynamics model of quadrotor aircraft are established.The cooperative path planning problem of multiple UAVs is decomposed into target assignment problem and local path planning problem,and the two subproblems are described mathematically.Secondly,the multi UAV target assignment problem is studied,and the centralized and distributed target assignment algorithms are proposed respectively for two different information constraints.By designing coding mode and crossover,mutation and selection operators,a target allocation method based on genetic algorithm is proposed;by combining the particle operation of genetic algorithm with the idea of particle swarm optimization,an improved particle swarm optimization target assignment algorithm is proposed.In view of the limited range of target perception and communication,a distributed target allocation method based on grid is proposed.The grid partition standard and intra grid matching algorithm are given.At the same time,the inter grid matching method is given by designing the communication mechanism between grids.Then,the local path planning problem of multiple UAVs is studied.By establishing the path planning process as Markov decision process,a path planning algorithm based on Multi-Agent Reinforcement Learning is proposed.By designing the state space and reward function,the planning problems of small-scale UAVs with and without assignment targets are solved.In order to extend the algorithm to large scale,a scalable and convergent distributed maddpg algorithm is proposed by introducing sharing strategy and local value function.In addition,aiming at the problems of low obstacle avoidance efficiency and acceleration not meeting the constraints of ORCA algorithm,an improved ORCA algorithm is proposed by speed preprocessing and virtual target point guidance.Finally,the practical application of the proposed path planning algorithm is studied.A set of multi aircraft experimental platform is built to deploy and analyze the algorithm.The feasibility and effectiveness of the algorithm are illustrated by experiments. |
