Project: Evolving Dynamic Traffic Assignment Heuristics.
A Thesis advised by Rutger Claes and Rinde Van Lon, and promoted by Tom Holvoet from the CS department at KU Leuven University.
This project combines the ECJ framework for Genetic Programming (http://cs.gmu.edu/~eclab/projects/ecj/), the wrapper developed by Rinde Van Lon and my java port of the DTA traffic simulator originally developed in MATLAB by Willem Himpe and other members of the Civil Engineering Traffic Group at KU Leuven.
The project consists on evolving the traffic re-assignment heuristic which is part of the algorithm used to attain dynamic user equilibrium, which in a few words, consists on ensuring that the simulated traffic matches realistic traffic behaviour (experienced drivers choose the fastest route).
Andres Arribas
(Master in Artificial Intelligence 2012/13)
andres.arribas AT gmail.com