laythamra28 / Deep-Q-Learning-Racing-Kart

Undergoing project to develop deep q learning project to teach kart to race

Deep-Q-Learning-Racing-Kart

Project Description

PySuperTuxKart is a racing game library that allows for sensorimotor control experiments. The objective of this project is to teach the agent, the kart, to drive around the tracks as fast as possible through deep Q reinforcement learning. As it currently stands the Kart seems to learn slowly after many episodes of traning however it can be improved. A full descitpion of the project is provided through this pdf: https://github.com/laythamra28/Deep-Q-Learning-Racing-Kart/blob/main/EC418%20Final%20Report.pdf

Installation

1. Clone this repo
2. Make sure all files are in the same directory when running
3. Run the main_practice.ipynb

Example showing the Q agent learning in early stage

__temp__.mp4