An environment for simulating encounters between aircraft for the purpose of creating collision avoidance systems. It is compatible with the collection of deep reinforcement learning models made available by OpenAI's baselines repo.
All of these packages are for Python, so make sure you have some version of Python 3 installed. I am running Python 3.7.4 for reference.
-
Install OpenAI Gym: Follow installation instructions here. TL;DR run
pip install gym. -
Install OpenAI Baselines: Instructions here. Requires that you have either Mac OS or Linux; you'll have to search around for instructions to run on Windows. Make sure you install TensorFlow with
pip install tensorfloworpip install tensorflow-gpuif you have an NVidia GPU.
Run the commands below to check if installation worked.
# Train model and save the results to cartpole_model.pkl
python -m baselines.run --alg=deepq --env=CartPole-v0 --save_path=./cartpole_model.pkl --num_timesteps=1e5
# Load the model saved in cartpole_model.pkl and visualize the learned policy
python -m baselines.run --alg=deepq --env=CartPole-v0 --load_path=./cartpole_model.pkl --num_timesteps=0 --play
The first command trains a DQN model and the cartpole environment. This step might take ~20 minutes if you're running it on a CPU.
The the second uses the learned state value function to try to balance the pole on top of the cart. You should see a window pop up with a 2D rendering where the DQN model is trying to balance a pole on top of a cart in real time.
After having installed gym, go to the repo directory and run pip install -e gym-ca. After this, you will be able to create the gridworld gym environment with gym.make('gym_ca:ca-gridworld-v0').