I built an open-source project that uses reinforcement learning to train and control a swarm of drones in Microsoft's AirSim simulator in Unreal Engine 4. It leverages Stable-Baselines3 for training policies and includes multi-agent custom environments with PettingZoo, and logging with Tensorboard.
GitHub: https://github.com/Lauqz/Drone-Swarm-RL-airsim-sb3Would love feedback from anyone into robotics, AI, or simulations; especially ideas on improving swarm coordination, generalization or scaling!