UAV Swarm Confrontation Based on Multi-Agent Soft Actor-Critic Method

The UAV swarm confrontation learning environment faces complex challenges due to its high dimensionality, nonlinearity, incomplete information, and continuous action space.To address these issues, this study proposes a multi-agent soft actor-critic (MASAC) deep reinforcement learning method based on incomplete information. Built on the centralized training-distributed execution (CTDE) framework,the proposed method establishes a UAV swarm confrontation game model and simulates a multi-UAV combat environment in continuous space. Simulation results demonstrate that the MASAC method outperforms existing multi-agent deep reinforcement learning techniques in terms of convergence speed and stability. The results of this study under score the practicality and effectiveness of the MASAC method in enabling intelligent decision-making for UAV swarms, thereby offering essential technical support for future advancements in UAV operations.