Windows folder structure (\) does not work on linux (/), and documentation?

[petterssonjonas]

Windows folder structure () does not work on linux (/). I had to replace all asset value strings with / for them to load. Should be an easy thing to fix.
If there is any documentation on what the different functions (keybinds) actually do, that would be cool.

Cool project, might put some time aside to hack on it to learn more ML.

[SevenChords]

i mean, you would either need a way to determine the OS during runtime to load the assets from different file paths or you would need different python files for different OS, but it is easily changeable yourself using any basic search and replace function

[minetonight]

Windows folder structure () does not work on linux (/). I had to replace all asset value strings with / for them to load. Should be an easy thing to fix. If there is any documentation on what the different functions (keybinds) actually do, that would be cool.

Cool project, might put some time aside to hack on it to learn more ML.

1. Generating an initial population of cars with random neural networks.
2. Learning as the game progress.
3. Applying a genetic algorithm to breed the next and better generation of cars.

Every generation consists of 200 cars, of which each have its own independent neural network.

The weights and biases for the first generation neural networks are assigned randomly.

Each neural network is formed by 3 layers:

1. An input layer with 5 neurons.
2. A hidden layer with 6 neurons.
3. An output layer with 4 neurons.

A car knows its distance to five points where it would crash (red lines) and uses his own neural network to perform an action:

0.5 = Accelerate
0.5= Brake
0.5= Turn Right
0.5 = Turn Left
Input Layer
Hidden Layer
Output Layer

In each generation, the fittest cars are selected and used to breed the next generation by using genetic algorithm operators:

1. Selection: This is done manually.
2. Crossover.
3. Mutation.

Selected cars are kept highlighted in green for easy comparison between them and their offsprings.

0-9- Change Mutation
LMB-Select Unselect
RMB-Delete
L-Show/Hide Lines
C-Clean
A-Toggle Player
R-Reset
B-Breed
N-Next Track
D-Toggle Info

Usage:
) select exactly 2 cars with the mouse
) press b - Breed
) run a few generations until a full lap.
) press n and see if the Neural network works on the second map too.
) optional - run mapGen.py and create another random map to test your cars.
) optional - press a and drive along the AI cars :-) get rect

Shortcuts
l - show lines for the Neural network
c - remove collided
m - mutate
r - reset all progress
n - new track
d - display info texts
a - autonomous car. You can steer a car with arrows
0-9 - rate of mutation 0-90