A self driving vehicle based on evolutionary neural network and reinforcement learning neural network
This program is mean to demonstrate how a self learning autonomous vehicle works, with either evolutionary algorithm or reinforcement learning. That means you can switch between these two methodes.
Start the program with evolutionary methode by running: python Game.py
If you want to start the program with reinforcement learning, run it with: python Game.py --rl True
You can try to change the game by editing the Config.py file, e.g. change the map size, change the car size etc.
Result of a evolutionary algorithm test:
This is the first generation
This is the last generation which has a individual fits the max required fitness
Rresult of reinfocement learning
This is one of the tests in the beginning:
This is one test that passed the max required fitness
There is a real-time statistics monitor for reinforcement learning, so you can see how is the current fitness is.
Have fun!