Evaluating Impact of Automated Vehicles on Operational Efficiency at Urban Intersections: Empirical and Simulation Study

With the growing deployment of automated vehicles (AVs) in urban road networks, their influence has become increasingly evident at intersections, where complex interactions with human-driven vehicles (HVs) can significantly affect traffic operations. While most studies have focused on freeway conditions, research in urban settings remains limited, often concentrating narrowly on start-up behaviors or crossing conflicts. This study analyzed the driving behaviors of AVs and HVs using trajectory data collected from 10 signalized intersections in Jiading District, Shanghai, China, and the Waymo Open Dataset in the US. Key metrics, including spacings, speeds, acceleration variability, and lane-changing time, were examined. The results showed that current operational AVs generally maintained larger spacings, traveled at higher speeds, and exhibited lower acceleration variability than HVs. These characteristics positively influenced surrounding HV behavior. An enhanced Intelligent Driver Model (E-IDM) was developed and calibrated with empirical data to capture car-following (CF) behaviors of both AVs and HVs in one unified framework. Based on the collected field data, the study further investigated AV impacts on traffic flow in upstream segments and intersection zones. It was found that AVs improved the speed and stability of following HVs in upstream areas but increased queue lengths and reduced speeds within intersections. These impacts differed across movement types (i.e., left-turn, straight-through, and right-turn). Simulation with the E-IDM further explored the influence of lane numbers, traffic demand levels, AV penetration rates, safety buffer distances, and time gaps. The results showed that in the upstream section, higher AV penetration rates and additional lanes strengthened the positive impact of AVs, whereas higher demand levels and time gaps aggravated it, and the safety buffer distances had little influence. In the intersection zones, higher AV penetration rates and additional lanes mitigated the negative impact, while higher demand levels and time gaps reduced it, and shorter buffer distances improved capacity. These insights offer practical guidance for optimizing AV deployment in urban traffic systems.