Impact of autonomous vehicles on the performance of a signalized intersection under different mixed traffic conditions: A simulation-based investigation

Abstract

Autonomous driving can overcome the limitations of stochastic human driving behavior. Therefore, implementing autonomous vehicles (AVs) could improve the efficiency of road networks. This study investigates the impacts of AV implementation on the performance of a signalized intersection considering a mixed traffic environment comprising regular vehicles (RVs) and AVs through microscopic traffic simulations. Accordingly, 24 scenarios with different AV implementation rates, AV driving models, and traffic volume conditions, were developed and evaluated using the Vissim simulation software. The results indicated that even partial AV implementation could improve the operational efficiency of a signalized intersection compared to full RV traffic. AV implementation reduced the vehicle delay, stopped delay, and queue length. The expected improvements are primarily based on the implementation rate, and are higher at higher rates (≥50%). The improvements are highest at moderate traffic volumes. Compared to the moderate level, partially replacing RVs with AVs at free-flow conditions does not significantly impact the performance of the intersection. Under congested conditions, the expected improvements from AV implementation are mitigated by the high traffic volumes. Considering the different AV models employed herein, the connected autonomous vehicle (CAV) model exhibited the best performance.