Analyzing the Impact of Autonomous Vehicles on Urban Traffic Flow at the Large Scale Network Using Real-World Data

This study evaluates the network performance of Daejeon City's traffic based on the market penetration rate of Autonomous Vehicles (AVs). Real-world AV data from the Waymo perception dataset is used to calibrate the parameters of a car-following model. AVs exhibit smoother speed changes, longer time headway, and a larger minimum gap compared to human-driven vehicles. These conservative driving behaviors enhance safety and comfort for passengers in AVs. And they result in a decrease in road capacity by approximately 35% when AVs reach 100% composition of the current road vehicle mix. The study validates previous assumptions and logic in AV research using real-world data, providing insights for policy-making and technological development. Future work should consider incorporating multiple datasets to gain a more comprehensive understanding of the impacts of AVs.