Evaluation of The Impact of Automated Driven Vehicles on Traffic Performance at Four-leg Signalized Intersections

Abstract: Intersections, particularly four-leg signalized intersections, are frequent sites of traffic congestion in urban areas. This congestion can lead to delays, increased travel time, and a negative impact on traffic performance and quality of life for people. Consequently, automated driven vehicles (AVs) have caught the attention of road operators and traffic engineers as these vehicles have the potential to enhance traffic performance by reducing traffic congestion. For example, AVs reduce traffic congestion and delays by controlling tasks done by human drivers, eliminating errors, and leading to smoother traffic flow . Based on this premise, this thesis focuses on evaluating the potential impacts of AVs on traffic performance at a four-leg signalized intersection . To account for the possible uncertainties that arise from the introduction of autonomous driven vehicles (AVs), this study employs a simulation approach to generate multiple scenarios that depict the potential evolution of AVs technology and the coexistence of different AVs. The scenarios contain three driving logics - cautious, normal, and all-knowing. Additionally, the study considers the impact of increasing levels of AVs penetration rates. Traffic performance is evaluated in terms of total travel time, average speed, average delay, and queue length. To do so, microscopic traffic simulation is used as it provides a safe and cost-efficient way to investigate the possible effects of AVs. The evaluation compares the traffic flow of three different driving logics of AVs (cautious, normal, and all-knowing) against human driven vehicles (HVs). PTV VISSIM software was used as a microscopic traffic simulation tool to analyze scenarios considering five penetrations rates 0% ,25%,50%,75, and 100% of AVs. Data from an existing four-leg signalized intersection in Norrkoping, Sweden, was used to build and calibrate the model in PTV VISSIM. The simulation results indicate that the driving logic of autonomous vehicles (AVs) impacts traffic performance at a signalized intersection. The cautious driving logic prioritizes safety, leading to slower traffic flow, longer queues, and increased delays. In contrast, the all-knowing driving logic prioritizes efficiency, which results in the most efficient traffic flow, higher average speeds, shorter queues, and reduced total travel time. Normal driving logic balances safety and efficiency, resulting in moderate improvements in traffic performance in contrast with other vehicles. Also results show that the penetration rate of AVs has a direct impact on traffic performance. A 50% penetration rate of all-knowing, normal, and mixed AVs with a 33% share percentage is necessary for significant improvements in total travel time, average speed, and maximum queue length. However, the average delay remains worse than that of human-driven vehicles until a market penetration rate of 75% for normal and mixed AVs with a 33% share percentage. Simulations reveal that cautious AVs negatively impact traffic performance, with 75% penetration leading to a sharp increase in total travel time. A minimum of 50% cautious AVs is needed to see clear reductions in queue length and average delay. However, average speed decreases significantly when the cautious AVs make up at least half of the traffic flow. In general, Cautious driving logic negatively impact traffic performance, while all-knowing driving logic improves it.