Design of MADDPG Capture Algorithm for Multiple UAV Cooperation

Based on the background of Multiple Unmanned Aerial Vehicle (Multi-UAV) game with limited attack and defense capabilities and communication range, a game model is established to study the capture and attack problem of a red UAV team against its blue target UAV in this paper. Under the Actor-Critic framework of multi-agent deep deterministic policy gradient (MADDPG) algorithm, the original MADDPG algorithm is improved according to the characteristics of the game scenario to solve the problem of UAVs’ number attenuation and sparse rewards. The simulation results shows that the algorithm designed in this paper has improved the convergence speed, learning efficiency and stability. The sub-networks make the algorithm more suitable for the game scenario where the number of UAVs is dynamically declining.