This is a fun project that I created to learn Reinforcement Learning, inspired by CodeBullet's awesome video on YouTube. It follows literally step-by-step according to his experiment, so maybe no creativity point for me 🐧 But still, it was a fun and educational experience.
- Tensorflow 2.0.0 or higher
- Keras 2.3.1 or higher
- Pyglet 1.4.8 or higher
python train_deep_q.py # run the Deep Q Learning's Training program. Press ESC to quit and save the model to file "model.h5"
python main.py # import model from "model.h5" to run the self-driving car
The repository includes a pretrained model.h5 that was created after about 40+ epochs, which is not much, but it can finish a full lap around the circus, which is something 🐧