This project is a road simulation with a ton of cars (almost 1,000). Each car has a multi perceptron neural network. The learning estrategy is evolutionary strategy. So, the cars run several generations and the best continue for others generations.
I did this project with python and cython, my inspiration was this video: https://www.youtube.com/watch?v=gnfkfUQvKDw&t=346s
To install this project is very simple! First, you need to run on your terminal:
git clone https://github.com/jonysalgado/automatic-cars.gitAfter, you need to compile some files, go to the project directory and run:
python setup.py build_ext --inplace
To run the simulation, you need to run:
python main.py
And that's it!
There are some keys that are useful for this project and I will list bellow:
- Z -> zoon in
- X -> zoon out
- A, D, S, W -> to move when you press Z
- F1 -> to save the best cars
The cars don't learn every road, but I think that improvements on the neural net or evolutionary strategy has to be done.
If you get any problem, please contact me: jonysalgadofilho@gmail.com
If you like to contribute, please do it! I will like so much, I did this project to help me to study AI and I think that can help you, as well.