North Carolina Department of Transportation
Research & Innovation Summit – 2021
Parametric Study of Car Following Model for Traffic Simulation using Genetic Algorithm
Authors: Matthew James Carroll, Rui Wu, and Jinkun Lee
East Carolina University
View or Download PDF
As the penetration rate of autonomous capable vehicles increases, traffic networks will go through a transitional period where auto-driving vehicles are introduced into traffic networks resulting in disruptions and safety risks. Since auto-driving vehicle behaviors can be controlled through an automated driving control system, there is a need to understand and control autodriving behaviors to maintain safety while allowing for optimal flow of vehicles. With this study we use a traffic simulation platform, Simulation of Urban Mobility (SUMO), to investigate the effects of SUMO’s autonomous vehicle car-following model, Intelligent Driver Model (IDM), using machine learning (ML) framework using genetic algorithm (GA) to determine the best IDM parameters, which include acceleration, deceleration, minimum gap, tau (relaxation time), and delta (acceleration exponent), to increase the flow of vehicles while maintaining low collisions for simulations with varying penetration rates.
For questions about this research, contact Matthew James Carroll at email@example.com.