Multi-UAV cooperative maneuver decision-making for pursuitevasion using improved MADRL  

作　　者：Delin Luo Zihao Fan Ziyi Yang Yang Xu 

机构地区：[1]School of Aerospace Engineering,Xiamen University,Xiamen 361102,China [2]School of Civil Aviation,Northwestern Polytechnical University,Xi'an 710072,China

出　　处：《Defence Technology（防务技术）》2024年第5期187-197,共11页Defence Technology

基　　金：supported in part by the National Key Laboratory of Air-based Information Perception and Fusion and the Aeronautical Science Foundation of China (Grant No. 20220001068001);National Natural Science Foundation of China (Grant No.61673327);Natural Science Basic Research Plan in Shaanxi Province,China (Grant No. 2023-JC-QN-0733);China IndustryUniversity-Research Innovation Foundation (Grant No. 2022IT188)。

摘　　要：Aiming at the problem of multi-UAV pursuit-evasion confrontation, a UAV cooperative maneuver method based on an improved multi-agent deep reinforcement learning(MADRL) is proposed. In this method, an improved Comm Net network based on a communication mechanism is introduced into a deep reinforcement learning algorithm to solve the multi-agent problem. A layer of gated recurrent unit(GRU) is added to the actor-network structure to remember historical environmental states. Subsequently,another GRU is designed as a communication channel in the Comm Net core network layer to refine communication information between UAVs. Finally, the simulation results of the algorithm in two sets of scenarios are given, and the results show that the method has good effectiveness and applicability.

关 键 词：Reinforcement learning UAV Maneuver decision GRU Cooperative control 

分 类 号：V279[航空宇航科学与技术—飞行器设计] E91[军事]