AWS Deepracer Simulation round solution
Solution was developed in collaboration with @DAOvb.
Solution is based on a simple PPO algorithm, achieves near-optimal score in AWS Deepracer environment.
To start training:
docker-compose up -d
python run_ppo_train.py
To create a gif of agent’s rollout:
docker-compose up -d
python create_gif.py save/best_416_iconic-field-98_6env_SCLight_4st1sk.pickle
