The Optimization of Traffic Signal Light using Genetic Algorithms coupled with a Simulation Tool (SUMO)

摘要

Maintaining a good Car Traffic system is a priority of the government. This paper presents a way for improving such kind of systems in crossroads, based on the use of Genetic Algorithm to optimize Traffic Light cycles characteristics. However, it is not easy to define an objective function as a mathematical expression; thus the use of the SUMO freeware can provide us traffic simulation in order to determine an objective function of a given traffic cycle. We define this function (referred to as "fitness function") as being the mean traveling time spent by the cars in the crossroad. Our algorithm gave us an improvement up to 20% in the cases we tried and has proven to be even more efficient for great traffic densities.