Research on UAV 3D Path Planning Based on Improved DQN
KONG Jianguo
ZHAO Tiantian
LIANG Haijun
LIU Chenyu
MA Kexin
Abstract：In order to solve the problems of poor convergence and low success rate of DQN in UAV path planning under un-known environment,a NoisyNet-DuelingDQN based pathplanning method is proposed.Based on the traditional DQN algorithm,the competition network is introduced to evaluate the value of each action better.Secondly,by introducing noise into the weight of the neural network,the space can be better explored and the optimal strategy can be found.Finally,the simulation results show that the algorithm hasbetter convergence and higher reward value than the traditional DQN and NoisyNet-DQN algorithms in different envi-ronments.After 60 000 simultaneous training times,the success rate of the algorithm is increased by 12.16%compared with DQN and 3.6%comparedwith NoisyNet-DQN.
Keywords：deep reinforcement learningpath planningDQN algorithmNoisyNet-DuelingDQN
Publication Date：2025-07-20
Online Publishing Date：2025-09-25(First online date of this platform, not the publication date of the document)
Pages：5( 36-40 )