Key dependencies based (Lower versions are not guaranteed to be feasible):

python: 3.9.0

numpy: 1.26.0

gym: 0.26.2

pillow: 10.2.0

torch: 2.4.0+cu124

torchaudio: 2.4.0+cu124

torchvision: 0.19.0+cu124

Explanation of Document

• agent/buffer/maddpg/networks: Refer to Phil's work -> PhilMADDPG;

• sim_env: Customized Multi-UAV round-up environment;

• main: Main loop to train agents;

• main_evaluate: Only rendering part is retained in main, in order to evaluate models (a set of feasible models is provided in tmp/maddpg/UAV_Round_up;

• math_tool: some math-relevent functions.