Dog Fighting is the summit of multi-UAV control problem. In the present work we extend the PPO algorithm to multi-UAV environment and investigate the decentralized learning of UAVs by MAPPO algorithm. By adding the resource baseline of UAVs in the classical rewarding scheme, the convergence of calculating the reward matrix is theoretically proven. We also demonstrate how the training of multi-agent PPO algorithm accelerates. Competitive UAVs learn to dog fight efficiently. Two groups of UAVs trained under resources baseline find an optimal strategy to keep the fighting as long as possible. The present work demonstrates that resource baseline MAPPO can become a practical tool for studying the decentralized learning of multi-UAV systems in highly complex environments.
Resource Baseline MAPPO for Multi-UAV Dog Fighting
Lect. Notes Electrical Eng.
International Conference on Autonomous Unmanned Systems ; 2021 ; Changsha, China September 24, 2021 - September 26, 2021
Proceedings of 2021 International Conference on Autonomous Unmanned Systems (ICAUS 2021) ; Chapter : 327 ; 3330-3336
2022-03-18
7 pages
Article/Chapter (Book)
Electronic Resource
English
Resource Baseline MAPPO for Multi-UAV Dog Fighting
TIBKAT | 2022
|Resource Baseline MAPPO for Multi-UAV Dog Fighting
British Library Conference Proceedings | 2022
|Trajectory Planning of Multiple Unmanned Aerial Vehicles on MAPPO
Springer Verlag | 2025
|Multi-UAV Path Planning for Air-Ground Relay Communication Based on Mix-Greedy MAPPO Algorithm
DOAJ | 2024
|