Assessing the influence of autonomous vehicles on traffic efficiency at signal-controlled intersections

In the context of evolving urban mobility, the integration of autonomous vehicles (AVs) holds significant promise and challenges, especially at signalized intersections, which are crucial for urban traffic management. This thesis investigates the complex dynamics of AVs at a signalized intersection in Balgat, Ankara, using microscopic traffic simulation with PTV VISSIM. The study encompasses 21 scenarios, varying between different AV driving behaviors (cautious, normal, aggressive, platoons, and mixed) and penetration rates ranging from 25% to 100%, in addition to human-driven scenarios. The findings reveal distinct impacts before and after signal control optimization. Before optimization, cautious AV behaviors, which prioritize safety, resulted in increased queue lengths, travel times, delays, emissions, fuel consumption, and conflict points, especially at higher penetration rates. Normal AV behaviors initially resembled human driving at 25% penetration but showed improvements in queue lengths, travel times, delays, and conflicts at higher penetration rates, despite slightly elevated emissions and fuel consumption. Aggressive AV behaviors significantly reduced queue lengths, travel times, delays, emissions, fuel consumption, and conflict points, achieving the most favorable results at 100% penetration. Platoon behaviors also reduced queue lengths, travel times, delays, and conflicts but indicate an increase in emissions and fuel consumption at 100% penetration due to the dynamics of platoon leadership. After optimization, significant improvements were observed across all AV behaviors for queue length, travel time, delays, emissions, and fuel consumption at cycle times of 60, 80, 100, 120, 140, 160, and 180 seconds. Human-driven scenarios at 0% AV penetration experienced substantial reductions in queue lengths, travel times, delays, and conflict points. Cautious AVs showed marked improvements across all penetration rates, although their values remained higher than other behaviors. Normal AVs significantly reduced all metrics with higher penetration rates after optimization. Aggressive AVs achieved the best outcomes across all metrics, particularly at 100% penetration. Platoon-forming AVs also demonstrated substantial improvements, though emissions and fuel consumption increased at 100% penetration due to the influence of the platoon leader. Notably, shorter cycle times showed better performance in terms of reducing queue lengths, travel times, delays, emissions, and fuel consumption compared to longer cycle times.The mixed behavior scenario consistently reduced queue lengths, travel times, delays, and conflicts across all penetration rates and exhibited trends in emissions and fuel consumption similar to human-driven scenarios with slight increases. Validation using field data ensured the accuracy of the model and provided practical insights for integrating AVs into existing traffic frameworks. This study highlights the transformative potential of AVs in enhancing urban traffic management and lays the groundwork for future research in AV integration and optimization.