Road traffic crashes are a major safety problem, with one of the leading factors in crashes being human error. Automated and connected vehicles (CAVs) that are equipped with Advanced Driver Assistance Systems (ADAS) are expected to reduce human error. In this paper, the Simulation of Urban MObility (SUMO) traffic simulator is used to investigate how CAVs impact road safety. In order to define the longitudinal behavior of Human Drive Vehicles (HDVs) and CAVs, car-following models, including the Krauss, the Intelligent Driver Model (IDM), and Cooperative Adaptive Cruise Control (CACC) car-following models were used to simulate CAVs. Surrogate safety measures were utilized to analyze CAVs’ safety impact using time-to-collision. Two case studies were evaluated: a signalized grid network that included nine intersections, and a second network consisting of an unsignalized intersection. The results demonstrate that CAVs could potentially reduce the number of conflicts based on each of the car following model simulations and the two case studies. A secondary finding of the research identified additional safety benefits of vehicles equipped with collision avoidance control, through the reduction in rear-end conflicts observed for the CACC car-following model.

Automated vehicles (AVs) will appear on the road soon and will influence the properties of road traffic networks such as capacity and safety. While the impact of AV on traffic operations has been discussed extensively, most existing literature in this direction focuses on microscopic and mesoscopic levels. This study examines the effect of AVs on traffic flow operations and crash risks at a network scale. In addition, this study discusses the implications for designing and operating safe, low‐speed urban road networks. Simulation experiments were carried out in a grid road network with AVs and human‐driven vehicles operating in mixed flow conditions. To assess traffic performance, the macroscopic fundamental diagram (MFD) relationships were estimated, specifically focusing on speed‐density correlations. From this analysis, it was possible to extract key traffic performance indices such as capacity and critical speed. Using time‐to‐collision surrogate safety measures, macroscopic safety diagrams were generated which associate the level of congestion with the potential crash conflicts among vehicles at an aggregated spatial scale. Utilizing this knowledge, a novel multiobjective optimization based on the NSGA‐II algorithm was applied to identify the optimal trade‐off between efficiency and safety. The presence of AVs was found to have a positive impact on the capacity, critical density, and average speed on a system level, even in low‐speed scenarios. Moreover, AVs can result in increased critical density in the network, which suggests that the road system can serve more vehicles at its capacity, thus improving efficiency, while decreasing the number of conflicts. These findings are useful for both traffic planners and operators.

Accurate traffic volume and speed prediction have a wide range of applications in transportation. It can result in useful and timely information for both travellers and transportation decision-makers. In this study, an Attention based Long Sort-Term Memory model (A-LSTM) is proposed to simultaneously predict traffic volume and speed in a critical rural road segmentation which connects Tehran to Chalus, the most tourist destination city in Iran. Moreover, this study compares the results of the A-LSTM model with the Long Short-Term Memory (LSTM) model. Both models show acceptable performance in predicting speed and flow. However, the A-LSTM model outperforms the LSTM in 5 and 15-minute intervals. In contrast, there is no meaningful difference between the two models for the 30-minute time interval. By comparing the performance of the models based on different time horizons, the 15-minute horizon model outperforms the others by reaching the lowest Mean Square Error (MSE) loss of 0.0032, followed by the 30 and 5-minutes horizons with 0.004 and 0.0051, respectively. In addition, this study compares the results of the models based on two transformations of temporal categorical input variables, one-hot or cyclic, for the 15-minute time interval. The results demonstrate that both LSTM and A-LSTM with cyclic feature encoding outperform those with one-hot feature encoding.

Urban signalized intersections face increasing pressures from congestion, emissions, and safety risks, yet most existing signal controllers remain predominantly platoon-blind and single-objective. In addition, among the limited studies that jointly address DRL signal control and platooning, many rely on macroscopic representations or simplified plugins that do not explicitly capture vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communication mechanisms. This study proposes MOPADRL, a multi-objective platoon-aware deep reinforcement learning (DRL) model for signalized intersection management in a microscopic V2X environment. In the proposed V2X framework, V2V communication enables automated vehicle platooning through coordinated gap regulation and string-stable car-following behavior, while V2I communication allows approaching platoons to transmit Platoon-Request messages to the signal controller. A platoon-augmented Double Deep Q-Network jointly optimizes waiting time, CO2 emissions, and platoon cohesion through an explicit severance penalty, and incorporates a conditional V2I-based Platoon-Request preemption channel that grants priority only under low-conflict conditions. Across 100-seed evaluations under both balanced and imbalanced demand profiles, MOPADRL, at platoon-size optima of seven vehicles for balanced demand and six vehicles for imbalanced demand, reduces queue length and waiting time by over 90% relative to fixed-time control and lowers CO2 emissions by up to 47% relative to human-driven baselines. The V2V platooning model eliminates time-to-collision conflicts across all demand scenarios. Microscopic trajectory analysis further validates the analytical gap design and confirms the string stability of the proposed platooning model under DRL-induced stop-and-go conditions.