Aircraft cluster air combat scenario is a long sequence decision-making task with complex state change, difficult control, which the use of general supervised learning, RNN network method is difficult to deal with, considering the excellent performance of multi-agent reinforcement learning in solving energy distribution, team games and other aspects in recent years, we established a set of six-degree-of-freedom aircraft simulation core of the 2v2 air combat scenario, proposed a multi-agent deep reinforcement learning agent based on the air combat conditions. The work introduced the basic concepts of deep reinforcement learning and the situation assessment involved in air combat, and trained a 6-degree-of-freedom aircraft model according to the Markov process of multiple agents. The final results show that the model generated by the training can generate a basic joint confrontation strategy, which has high research value and practical significance.
2v2 Air Combat Confrontation Strategy Based on Reinforcement Learning
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2022 ; Xi'an, China September 23, 2022 - September 25, 2022
Proceedings of 2022 International Conference on Autonomous Unmanned Systems (ICAUS 2022) ; Chapter : 125 ; 1368-1377
2023-03-10
10 pages
Article/Chapter (Book)
Electronic Resource
English
Decision-making and confrontation in close-range air combat based on reinforcement learning
Elsevier | 2025
|UAV Swarm Confrontation Using Hierarchical Multiagent Reinforcement Learning
DOAJ | 2021
|