Comparative conflict analysis of human-driven and autonomous vehicles at signalized intersections

Integrating autonomous vehicles (AVs) into urban traffic systems presents both opportunities and challenges, especially at signalized intersections. This study offers a comparative conflict analysis of human-driven vehicles and AVs at a busy four-legged signalized intersection in Balgat, Ankara, Turkey. Using PTV VISSIM for detailed traffic simulation, the research assesses the effects of various AV driving styles - cautious, normal, aggressive, and a mix of all three - at different penetration rates (25% to 100%), alongside standard human-driven vehicle scenarios. The Surrogate Safety Assessment Model (SSAM) is employed to analyze safety implications both before and after intersection design calibration. The findings demonstrate notable differences in conflict points between human-driven and AV scenarios. Before calibration, cautious AV behaviors result in higher conflict points due to increased queuing, while aggressive behaviors reduce conflicts through more efficient traffic flow. Human-driven vehicles exhibit varied conflict levels based on driver behavior. After calibration, significant improvements are observed across all scenarios, with aggressive AVs achieving the greatest reduction in conflict points. This study highlights the potential for AVs to improve intersection safety and efficiency when appropriate design calibration measures are implemented.