ONLY WORKS ON LINUX

First you need to generate the worlds

run deno --allow-read --allow-write ./generateWorlds.ts

To start training

run python3 trainerStart.py [NUMBER OF ENVIRONMENTS] [NUMBER OF ROBOTS PER ENVIRONMENT] [Continue training? (true or false)]

or to start in debug mode run python3 trainerStart.py true this will start one visual training environment with 2 robots

Inference

run python3 inferenceStart.py

The neural network (notes for myself)

{ 18 motor pos 3 angle 3 angular acc 3 lin acc

2 turn rate, walk speed 1 turn rate walk speed mask

18 target pos 18 target pos mask }(66) * 3 (3 older) = 198