My task was to design and implement a genetic algorithm-based method to optimise the elements of a traffic network. The aim of the task was to optimise the networks for certain goals using a traffic modelling multi-agent system.
I have designed the genetic representation of a traffic network which describes the attributes of the network elements, like the roads and intersections. I have designed some modifications and extensions to the mentioned traffic simulator in order to use it in a genetic algorithm. I have considered how to rate a traffic network and designed fitness functions for the genetic algorithm.
The planned improvements were implemented to the simulator using the Eclipse-EJADE environment. I have implemented the genetic algorithm with the help of the JGAP framework. Finally, I have fixed the remaining stability issues in the software.
After finishing the software, I have run several simulations. I used different traffic networks to optimise and tested all the fitness functions, and finally I evaluated the results.