Code created for bachelor degree diploma project. The goal was to create a line follower robot steered by artificial neural network.
The neural network was trained using Deep Q-learning algorithm in a custom simulation environment created with PyGame.
Then it was extracted to physical robot that uses Raspberry Pi as main computing unit.
You don't really have to worry about it if you're using pipenv (and if you aren't you really should be).
Just run
pipenv install
to create virtual environment with all the required packages listed in Pipfile.
Then start it using
pipenv shell
and now your development environment is ready.
For running tests make sure you have RPi.GPIO and TensorFlow 2.3.0 installed on your Raspberry Pi.
As installation of TF 2.* is not that simple on older models of RPi, I recommend reading this guide to do so.