A small neat ai learning how to play chromium t-rex runner
NeuroEvolution of Augmenting Topologies is a genetic algorithm
for the generation of evolving artificial neural networks developed by Ken Stanley in 2002.
I started with a population of 25 t-rex, fitness for each individual is equal to score.
Fitness threshold is equal to 10 so whenever an individual reaches score 10, the simulation runs up
to the point where all t-rex are extinguished and doesn't repeat itself. After that, a winner.pkl file
is created with the champion genome.
It is used one parameter as input which is the distance between t-rex and next cactus and neat generates
two outputs which are used as criteria to jump or not.
python 3.8
install requirements: "pip3 install -r requirements"
run: "python3 game.py"
Assets for this game can be found Here!