siddrrsh / rl_airport

Multi-agent Airport RL

Usage:

You can run an existing multi-agent algorithm via weights and biases below.

python minimal-marl/idqn.py --env ma_gym:Airport4-v0

python minimal-marl/qmix.py --env ma_gym:Airport4-v0

python minimal-marl/vdn.py --env ma_gym:Airport4-v0

python minimal-marl/maddpg.py --env ma_gym:Airport4-v0

You can also run the environment with the default multi-agent backprop:

python examples/random_agent.py --episodes 1000 --env Airport4-v0

Demo:

openaigym.video.0.20412.video000001.mp4

Setup and Testing:

• Install: pip install -e ".[test]"
• Run: pytest