Learning to drive using neural networks.
This repository contains three subdirectories:
- plan/ for the plan
- doc/ for the final documentation
- src/ for all the source code
The project is a reinforcement learning implementation done during the C++ programming course held at Aalto University. It involves running a physics-based car through a racetrack. The car is controlled by a neural network, which then learns how to complete laps on the track as fast as possible. The project is far from perfect, but was a valuable learning experience.
The Box2D library is built for 64-bit systems, but it shouldn't be a problem.
=======
qlearning2
Team Skele aihe
Esa Koskinen
Simo Muraja
Olli Kauppinen
Jussi Hirvonen