Simulation of Genetic Algorithm: Traffic Light Efficiency

TL;DR

This paper presents a genetic algorithm-based approach to optimize traffic light schedules, improving traffic flow efficiency in urban intersections through simulation and iterative enhancement.

Contribution

It introduces a novel genetic algorithm framework integrated with SUMO simulation to optimize traffic light timing, outperforming traditional static schedules.

Findings

Genetic algorithm improved traffic flow efficiency.

Enhanced scheduling reduced total vehicle exit time.

Simulation results showed better performance than static schedules.

Abstract

Traffic is a problem in many urban areas worldwide. Traffic flow is dictated by certain devices such as traffic lights. The traffic lights signal when each lane is able to pass through the intersection. Often, static schedules interfere with ideal traffic flow. The purpose of this project was to find a way to make intersections controlled with traffic lights more efficient. This goal was accomplished through the creation of a genetic algorithm, which enhances an input algorithm through genetic principles to produce the fittest algorithm. The program was comprised of two major elements: coding in Java and coding in Simulation of Urban Mobility (SUMO), which is an environment that simulates real traffic. The Java code called upon the SUMO simulation via a command prompt which ran the simulation, received the output, altered the algorithm, and looped. The SUMO component initialized a simulation in which a 1 x 1 street layout was created, each intersection with its own traffic light. Each loop enhanced the input algorithm by altering the scheduling string (dictates the light changes). After the looped simulations were executed, the data was then analyzed. This was accomplished by creating an algorithm based upon regular practice, timed traffic lights, and comparing the output which was comprised of the total time it took for all vehicles to exit the system and the average time it took each individual vehicle to exit the system. These different variables: the time it took the average vehicle to exit the system and total time for all vehicles to exit the system, where then graphed together to provide a visual aid. The genetic algorithm did improve traffic light and traffic flow efficiency in comparison to traditional scheduling methods.