With the development of military informatization, military evolution is bound to include a large number of confrontations with artificial intelligence algorithms. In this paper, we discuss the air-to-air combat problem in that both sides have multiple UAVs involved, and propose a UAV Cluster Dominance Decision Algorithm Based on Reinforcement Learning algorithm (UCDD-RL). The combat dominance of each UAV is evaluated according to the live number, closeness, health, and relative positions of both camps in the local UAV cluster. The combat dominance of the UAV is the input of the fuzzy inference system, and its output is the UAV’s combat strategy. The combat strategies include fight, escape, occupy, and change cluster. We perform numerical simulations with the UCDD-RL algorithm in the UAV combat environment. By analyzing the reward, the win rate, and the live number of UAVs in the experiments, we verify that the UCDD-RL algorithm has a positive effect on improving UAV air combat level.
Reinforcement Learning-Based Cooperative Adversarial Algorithm for UAV Cluster
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 : 102 ; 1129-1138
2023-03-10
10 pages
Article/Chapter (Book)
Electronic Resource
English
Environment Adversarial Reinforcement Learning
AIAA | 2024
|A Cluster-Based Cooperative Localization Algorithm
IEEE | 2015
|CERTIFIED ADVERSARIAL ROBUSTNESS FOR DEEP REINFORCEMENT LEARNING
European Patent Office | 2021
|