bluesky_gym_population_density_extension

This dataset contains the trained policies as neural network weights and simulated trajectories as csvs, produced by a reinforcement learning agent that balances noise and fuel using population density observations processed by a CNN. It accompanies the Master's thesis "Balancing Noise and Fuel: Spatially Aware Reinforcement Learning for Air Traffic Control," and includes the source code (a fork of BlueSky-Gym).