Research And Implementation Of Path Planning Algorithm For Multi UAV Based On Reinforcement Learning

UAV technology has become the most important strategic goal of all countries.Because of its small size,convenience,low cost,it is widely used in the military and civil fields.UAV technology is also transiting from single UAV to multi UAV cooperative work,so it is very important to study multi UAV cooperative control.At present,the research of UAV task planning mainly focuses on the traditional a * algorithm,artificial potential field method,D * algorithm and some early popular intelligent algorithms,such as ant colony algorithm,particle swarm algorithm,genetic algorithm,etc.For single UAV,the running time of these algorithms is acceptable.However,in the case of multiple UAVs working together,the running time of the above algorithm will exponentially increase,and it is difficult to meet the time requirements in emergency.In view of these shortcomings,this paper designs a multi UAV path planning algorithm based on reinforcement learning for different business scenarios,and implements the system.Firstly,this paper introduces some basic knowledge of reinforcement learning,and then designs a single UAV single objective path planning algorithm,and optimizes it by initializing Q value table in advance and compressing state space,so that the operation efficiency can be significantly improved.Secondly,according to the situation of single UAV executing multi-objective,we design a path planning algorithm,and analyze the shortcomings and shortcomings of this scenario.Since then,based on maddpg algorithm and neural network,this paper designs a path planning algorithm for multiple UAVs,which makes UAVs use experience playback mechanism to learn neural network,so as to reduce the time to reach the target and avoid the collision between UAVs.The simulation results show that the algorithm is suitable for different environments and does not need repeated training.Compared with the traditional algorithm,the efficiency of the algorithm is significantly improved.Finally,on the basis of the above algorithm,the web site is designed with nginx,Java,My SQL and other technologies to realize the online execution of clustering algorithm and multi UAV path planning algorithm.