This repository is created for the project related to the "Natural Computation" subject at the University of Salerno.
The aim of the project is to learn, or otherwise design, a controller which could be able to race for a certain number of laps on a set of unknown tracks against other controllers. In particular, the primary goal for the project is to design a controller that runs the circuits in the shortest possible time without causing damage and arriving at the first places.
It is necessary to use the APIs and the Snakeoil Controller to develop a solution. And also it is necessary to use TORCS: The Open Racing Car Simulator TORCS has often used as a platform for the development and improvement of intelligent agents for autonomous driving in a competitive environment. Moreover, is important to choose an algorithm from those studied during the course and to define the fitness function taking into account the performance of the controller.
The constraints are:
- the number of laps should be at least 2;
- the number of opponents in a race should be at least 8 and should include: damned 1, damned 3, damned 4, damned 5, damned 6, berniw hist 2, berniw hist 6, berniw hist 7;
- the number of circuits where to train your controller should be at least 2 among Forza, CGTrack-2, E-Track-3, Wheel-1.
For more information, read the report.