ABSTRACT. Infrastructure managers are increasingly faced with challenges in the (re)designing and operation of railway stations as multimodal and multiservice hubs, particularly in relation to overcrowding and congestion, which result in difficult and conflicting pedestrian flows. Therefore, understanding how users access and leave the station (the transport mode they utilize and the station entrances they choose) as well as their movement within the premises, becomes essential for effective station design and management. This necessitates innovative tools that advance current practices. In this context, this paper presents an integrated macroscopic and microscopic modelling framework developed for the design and management of railway stations as multimodal hubs, with an application to the Milano Bovisa Politecnico railway station, a key transport hub in the municipality of Milan. This application enables the assessment of the station’s current performance and validates the proposed approach, including assumptions related to the macroscopic model and the microsimulation of passenger flows within the station. This research forms part of Spoke 9, Task 2.2 of the MOST project and was conducted in collaboration with the infrastructure manager Ferrovienord (FN).

ABSTRACT. Accurate traffic forecasting is essential for optimizing traffic management and transportation network efficiency. Traditional deep learning models often require large datasets and extensive training, limiting their adaptability to dynamic traffic conditions. Moreover, in the presence of limited data, the prediction task becomes even more overwhelming. In order to overcome these challenges, this study proposes a novel approach leveraging Model-Agnostic Meta-Learning to optimize a Long Short-Term Memory network for link flow prediction. In fact, the Model-Agnostic Meta-Learning framework enables the Long Short-Term Memory model to rapidly adapt to new traffic patterns with minimal data, improving prediction accuracy and robustness. The aim of the proposed method is to obtain a generalized model that, after slight fine-tuning, it is able to make predictions in new demand or spatial contexts without the need for numerous data. The application to a real network in the city of Rome, Italy, with simulated data demonstrated its superior performance over conventional LSTM models in terms of generalization and forecast accuracy. The results highlight the potential of meta-learning for enhancing predictive capabilities in intelligent transportation systems.

ABSTRACT. The paper presents a digital platform for the management of online urban mobility that applies a hybrid rolling horizon and Bayesian filtering approach to traffic prediction based on dynamic traffic assignment and a mesoscopic simulation model. Several tests of consistency and efficiency show that the performances increase linearly with the extent of the simulation and less than linearly with the timestep of the simulation. The analyses conducted through real-size networks have also shown the need for an automatic simplifying procedure of graphs that eliminates all the nodes connecting only two links and joins the two links.

ABSTRACT. Urban and industrial areas increasingly face complex mobility challenges due to growing commuter demand, infrastructure constraints, and the need for sustainable transport solutions. However, there remains a lack of advanced digital tools to support long-term planning and scenario-based decision-making in such contexts. This paper addresses this gap by presenting BISTWIN, a Mobility Planning Digital Twin (MPDT) designed to support mobility planning in the Bissen industrial zone, Luxembourg. BISTWIN integrates historical traffic data with a high-performance simulation engine to evaluate the impact of mobility interventions. The key contributions include: (i) developing an MPDT tailored to the Bissen industrial zone, (ii) validating the model with publicly available traffic data, and (iii) demonstrating its decision support capabilities through two “what-if” scenarios. BISTWIN offers a scalable framework for sustainable transportation planning, providing valuable insights for future MPDT applications in industrial and urban settings.

ABSTRACT. This paper presents a novel approach to traffic management in road networks, consisting in time-varying class-specific variable speed limit (VSL) restricted to a fraction of road users. In particular, we present a macroscopic approach where traffic dynamics is described by a multi-class Lighthill-Whitham-Richards (LWR) model, with two classes of users (controlled and uncontrolled vehicles). The model can be applied to a general road network and our goal is to optimize traffic performance by minimizing both the average travel time in the network and the total time spent in virtual buffers at the network entries to prevent spill-back scenarios. The optimization is performed by acting on different ratios of controlled vehicles, and we compare the performance of the proposed control strategy with a classical inflow control at the entries of the network. The numerical tests show that class-specific speed control outperforms inflow control, and highlight the importance of tailored traffic control strategies for road networks, offering insights into optimizing mobility, safety, and traffic efficiency.

ABSTRACT. Traffic congestion in urban areas, especially during rush hours, is a major socio-economical concern that has a large impact in many areas. Intelligent traffic routing can mitigate the issue by routing traffic such that it better utilises the road network and make it less prone to bottlenecks that cause congestion. Centralised traffic routing techniques take a holistic perspective of the traffic situation in a region and optimize the traffic for given global criteria.

This paper concerns the online application of centralised traffic routing methods -- one based on Automated Planning and the other based on Mixed-Integer programming -- in larger metropolitan areas. In particular, centralised routing techniques are applied in smaller regions that are prone to congestion in given time intervals. To do so, it is important to estimate which vehicles arrive in the region in a given time interval, so they can be routed. We evaluate our approach on real historical datasets from Dublin and Luxembourg.

ABSTRACT. Urban traffic congestion significantly impacts air quality and contributes substantially to pollutant emissions. Effective traffic management strategies, therefore, require models that accurately capture both traffic dynamics and associated emissions. This paper proposes a Model Predictive Control (MPC) framework aimed at emission reduction in integrated urban-freeway networks. The traffic network is partitioned into an urban core, modeled via a Macroscopic Fundamental Diagram (MFD), and a peripheral freeway, where traffic flow is captured using a Bounded Acceleration extension of the classical Lighthill-Whitham-Richards (BA-LWR) model. The BA-LWR model enhances behavioral realism by incorporating driver responses to congestion, notably the phenomenon of capacity drop. Control measures include variable speed limits (VSL) and ramp metering, which are dynamically optimized through MPC to balance traffic efficiency and emissions minimization. Emissions are quantified using the COPERT III model, while route choice behavior between urban and freeway routes is modeled probabilistically based on real-time travel conditions. Simulations illustrate the framework’s capability to manage congestion and emissions effectively under variable traffic demand scenarios.

ABSTRACT. This study presents a backcasting approach that considers the passenger car fleet dynamics to determine optimal policy roadmaps in transport systems. As opposed to the scenario-based approach, backcasting sets emission reduction targets first, then identifies policies that meet the constraint. The policy is the implementation of Low Emission Zones (LEZs), in the ˆIle-de-France region as a case study. The aim is to minimize the number of scrapped vehicles due to LEZs under CO2 emission targets and to deduce an interdiction schedule of polluting vehicles by 2050. To explore potential solutions, we use a genetic algorithm that provides a first insight into optimal policy pathways.

ABSTRACT. Road transport emissions significantly contribute to urban air pollution, necessitating effective strategies to mitigate their environmental impact. Traditional routing typically optimizes travel time or distance, neglecting emission considerations. This paper proposes an eco-routing approach integrating emission calculations as a cost factor into path selection. Emissions for each road segment are estimated using the COPERT model, and Dijkstra’s algorithm is applied to compute three routing scenarios: (1) shortest travel-time path, (2) lowest-emission path, and (3) a balanced path integrating both travel time and emissions through monetary valuation. Results indicate that optimizing for the shortest travel time does not consistently yield the most environmentally friendly path. Specifically, routing based on lowest emissions achieved approximately a 29.78% reduction in emissions compared to the fastest path, although it increased average travel time by 19.86%. These findings underscore the importance and effectiveness of incorporating emission factors into routing decisions, highlighting their potential to significantly enhance sustainable transportation planning and urban air quality improvement.

ABSTRACT. Fixed time strategy is a common approach in signal traffic control in which signal plans are simple and periodic, enjoying easy implementation without detection mechanisms. A traffic light is associated with several daily plans, each applied to several consecutive hours. Time-of-day breakpoints (TODs) refer to the times over the day in which the plan is changed. TODs are often selected based on traffic, aiming to divide the day into groups of consecutive hours with similar traffic characteristics within each group of hours. We present a methodology to study time-of-day breakpoints in practice. We use this methodology to estimate and analyze time-of-day breakpoints in the city of Rio de Janeiro, Brazil based on traffic properties derived from traffic trajectories. Our study examines over 900 of the city intersections. We refer to properties such as the number of daily plans and the times by which plans start. We also provide traffic-aware insights on the potential improvement in the selection of TODs and identify key intersections where adjusting TODs could reduce average delay times. We identify potential improvements in over 8% of the examined intersections. These findings provide valuable insights for traffic engineers seeking to optimize signal timing.

ABSTRACT. Urban traffic congestion is a growing challenge, and optimizing signal timing strategies is crucial for improving traffic flow and reducing emissions. The coordination of signalized intersections improves both traffic operations and environmental aspects. Coordination is particularly important along arterials, sequences of signalized intersections that serve as the primary routes and carry a high volume of traffic. In this paper we analyze real data from the city of Rio de Janeiro to study properties of arterials. We refer to their length, the distance between intersections and to the properties of the traffic light plans such as cycle time. We then study their coordination level in practice in terms of number of stops and their common locations along the arterials. We dive into particular arterials and provide insights that can be useful for efficient design of arterials in additional cities. Based on the analysis, we show how simple traffic properties can indicate the potential upon coordinating two adjacent intersections as part of an arterial in improving traffic performance.

ABSTRACT. Optimizing signal timing in traffic lights helps to improve traffic flow and reduce emissions through reducing delay time in intersections. At intersections, vehicles from different movements observe different delays impacted by the traffic light plan. This paper analyzes delay fairness among various vehicles at intersections. We refer to three cities: Rio de Janeiro, Hamburg and Seattle with a total number of over 5100 intersections. We present an intuitive methodology to compute delay fairness based on Gini index, a common fairness measure in Economics. We evaluate the fairness based on real traffic data and provide insights on the relationship of fairness with day hours and traffic demand. We also examine real changes in traffic light plans that occurred in practice to check whether improving delay is often aligned also with increasing fairness.

ABSTRACT. The convergence of deep reinforcement learning (DRL) and advanced sensor technologies is transforming Traffic Signal Control (TSC), enhancing its adaptability and efficiency. Priority vehicles (PVs) are pivotal in safeguarding public safety and enhancing emergency response, so their efficient passage is of utmost importance. In the context of Vehicle-Road-Cloud Integration (VRCI), real-time data interaction and collaborative processing can be realized among road infrastructure, vehicles, and the cloud. This trend offers us an opportunity to capitalize on their advantages and better optimize the guidance of priority vehicles. Therefore, this paper presents a method namely Cloud-enabled Co-optimization of Priority Vehicle Preemption and Traffic Signal Control (CCPVLight). This method dynamically fuses the sensing data from road, the status of vehicles, and the computing from the cloud to jointly optimize the priority strategy on TSC. By designing a priority module to extract the phase priority vector and adopting a multi-objective decision-making mechanism to optimize the phase selection scheme, it significantly enhances the adaptability to complex and dynamic mixed traffic flows. Comprehensive experiments were conducted using the Simulation of Urban Mobility (SUMO) simulator. The results from both the training and testing scenarios have proven the effectiveness of CCPVLight, indicating its potential for real-world applications.

ABSTRACT. In the era of 100% connected and automated vehicles (CAVs), vehicles will be controlled without traffic signals at intersections, using the so-called Autonomous Intersection Management (AIM). Most research on AIM so far has focused on vehicle-related benefits such as lower average vehicle delay or higher vehicle throughput. This paper is the first to compare AIM at different intersection layouts and to investigate lane removal options to reallocate urban road space in the CAV era for active modes and public realm. For this purpose, we simulated a conflict-point and optimization-based AIM with integrated pedestrian crossings using the microscopic simulation tool SUMO for a variety of scenarios and compared it with original traffic-actuated signal controls (TSC) from Ingolstadt, Germany. Various pedestrian and vehicle demand levels and patterns (synthetic and real ones) were examined. To demonstrate the great potential of AIM in gaining space, suggestions for the redesign of a real intersection leg are made, including wider bike lanes, sidewalks, flex zones, and more recreational space. The results show that AIM allows for the removal of vehicle lanes, with the number of lanes removed depending on the level of pedestrian demand and the symmetry of vehicle demand. In a case study, we demonstrate that today’s signalized intersection with moderate pedestrian demand and two lanes in each direction plus a turn lane can be replaced with an AIM controlled intersection with one lane in each direction, while maintaining the current level of service for vehicles (LOS C) and pedestrians (LOS B).

ABSTRACT. With the rapid development of automated and connected vehicles, we see lane-free traffic as an emerging alternative to conventional lane-based traffic. Lane-free traffic should, in theory, eliminate the need for parallel lanes and allow vehicles to use the full width of the road more effectively, significantly increasing road capacity. However, the impact of increased traffic volumes and different traffic patterns on the serviceability of infrastructure, particularly highway viaducts, is still unknown. The objective of this study is to investigate the effect of lane-free traffic on the aging of viaduct pavements by comparing it with conventional lane-based traffic. A vehicle-bridge interaction simulation framework is used to simulate the dynamic traffic loads on a bridge structure under realistic lane-based and simulated lane-free traffic cases. The results indicate that although lane-free traffic results in heavier traffic on the same viaduct section compared to conventional lane-based traffic, even the simplest control strategy can significantly slow pavement fatigue through load redistribution. These results suggest that with broadly adapted lateral control algorithms that include vehicle-bridge interaction in a feedback loop, an AI-enhanced intelligent traffic system should be able to further optimize both mobility efficiency and infrastructure sustainability.

ABSTRACT. The Combined Alternate-Direction Lane Assignment and Reservation-Based Intersection Control (CADLARIC) is a novel approach designed to optimize urban traffic flow by managing directionally unrestricted traffic to improve efficiency while reducing the number of conflicts. Leveraging connected and automated vehicles, CADLARIC allows vehicles to utilize lanes typically reserved for the opposite direction. While previous research has focused primarily on vehicular flows, this study extends CADLARIC’s applicability to bicycle operations in an autonomous intersection environment. Specifically, we introduce a reservation-based method for serving bicycles, comparing two distinct reservation algorithms: the Priority-Based Algorithm and the Delay-Based Algorithm. The performance of CADLARIC is evaluated against Fixed-Time Control (FTC) as a baseline and Fully Reservation-Based intersection Control (FRIC) with conventional lane assignments. The results indicate that serving bicycles without granting them maximum priority allocates more time for vehicle movement, thereby enhancing CADLARIC’s overall performance in multimodal traffic conditions. These findings suggest that integrating bicycle reservation mechanisms within autonomous intersection environment can significantly improve efficiency and scalability in autonomous traffic management.

ABSTRACT. Connected and Automated Vehicles (CAVs) are a technological advancement that can transform the current mobility system. These vehicles, in addition to being more efficient than traditional ones, can be used to implement vehicle-based control strategies, as the aim of this work. Specifically, this paper presents a freeway control strategy where the speed to be maintained along the traffic stretch is actuated by groups of CAVs, called clusters. Differently from previous approaches, this study proposes a Distributed Deep Q-Network (D-DQN) based on Multi-Agent Reinforcement Learning (MARL) theory to determine the speed that CAV clusters should maintain in a certain portion of the freeway to mitigate congestion. In this regard, an extended version of the Cell Transmission Model (CTM) is leveraged to emulate the traffic dynamics with the presence of CAV clusters. A Numerical analysis, considering a stretch of the A20 freeway in the Netherlands, compares the proposed approach with the centralized version of the control scheme, demonstrating the effectiveness and the advantages of the proposed strategy.

ABSTRACT. Teleoperated driving enables remote human intervention in autonomous vehicles, addressing challenges in complex driving environments. However, its effectiveness depends on ultra-low latency, high-reliability communication. This paper evaluates teleoperated driving over 5G networks, analyzing key performance metrics such as glass-to-glass (G2G) latency, RTT and overall steering command delay. Using a real-world testbed with a Kia Soul EV and a remote teleoperation platform, we assess the feasibility and limitations of 5G-enabled teleoperated driving. Our system achieved an average G2G latency of 202.41ms and an RTT of 46.63 highlighting the G2G latency as the critical bottlenecks. The steering control proved to be mostly accurate and responsive. This paper provides recommendations and outlines future work to improve future teleoperated driving deployments for safer and more reliable autonomous mobility.

ABSTRACT. Over half of motor vehicle collisions resulting in injuries or fatalities occur at intersections. Rapid emergency response is crucial for saving lives. However, ambulances are often randomly positioned. When a car accident occurs at an intersection, the nearest ambulance is dispatched. This work presents a dynamic optimization model for ambulance placement, incorporating environmental conditions and historical accident data to enhance response efficiency. The model was evaluated through a traffic simulation of Detroit, demonstrating significant improvements in response times compared to published data.

ABSTRACT. Traffic incidents on roads, defined as non-recurring events causing a temporary reduction in capacity, causes congestion, injuries and public costs. Knowledge about how travelers change their behavior, in particular how they adjust their route, during incidents can give important insights into how to manage traffic when incidents occur. This paper aims to analyze the impact of incidents on route choice to enable decision support systems during incidents.

Reported incidents and GPS trajectories from road vehicles in Stockholm are used in the analysis. We propose a method to map incidents to network links using speed data. We also analyze incident impact on speed reduction, route choice at the network level, and route choice at the local level. A route choice model based on GPS data is evaluated on incident scenarios.

The result shows that 5-12 % of the incidents have a significant impact on the route choice. The impact of an incident on route choice appears to be influenced by the type and location of the incident, its duration, and the extent of the resulting speed reduction. Up to 22 % of the travelers change their route to alternatives not passing the incident. The route choice model tends to overestimate this number, predicting 43-100 % of the travelers to change their route.

ABSTRACT. The number of traffic accidents has been steadily increasing in recent years, and pedestrians are the most vulnerable road users. In response to the increasing mortality rate of pedestrians in car accidents, the influencing factors affecting the severity of pedestrian injuries were investigated, and the most significant influencing factors were compared and analyzed using both ordered probit models and multinomial logit models. The article obtained traffic crash data from 2013-2017 in North Carolina, USA, and selected 18 factors as independent variables from six aspects: pedestrian characteristics, driver characteristics, vehicle characteristics, location and roadway characteristics, environmental and temporal characteristics, and traffic control type. There are five levels of severity dependent variables: pedestrian uninjured, slightly injured, seriously injured but not disabled, seriously injured and disabled, and fatally injured. The analysis of the two models showed that light conditions, 22:00-05:59, maximum speed limit, and vehicle type were the common significant variables for both models at the 0.05 significance level and were significantly associated with the severity of pedestrian injuries. Finally, based on the results of the study, targeted measures to prevent pedestrian accidents are proposed, which also hope to provide data support and theoretical basis for traffic managers.

ABSTRACT. Ensuring road safety is a fundamental challenge in transportation systems, particularly on roads with high traffic volumes and varying geometric features. This study presents a general road safety analysis framework that exploits different types of data on traffic, geometry, and accidents to identify the portions of the network to be enforced with safety measures and potentially support drivers with an advanced onboard speed advisory system. The objective is to identify and classify homogeneous road elements to individuate the portions of the road network that need additional safety measures and that have to be indicated to drivers according to their driving attitude and risk perception. Using Floating Car Data (FCD), road geometry data, and historical accident reports, multiple clustering tests were conducted to group road segments where factors such as curvature, speed variability, and risk indicators contribute to increased accident probability. The proposed framework is applied on Via Pontina, a critical and strongly congested rural road in Italy, and among several clustering tests performed, four were selected based on their interpretability. The results illustrate the effectiveness of a data-driven approach for road safety assessment, offering valuable insights for transportation engineers and policymakers.

ABSTRACT. The Intelligent Video Analytics (IVA) project, backed by the Lazio Region, harnesses advanced Artificial Intelligence (AI) and 5G technologies to enhance traffic safety through real-time infrastructure-to-vehicle (I2V) communication. Utilizing real-time video processing, the IVA system identifies roadway anomalies such as accidents, rapid queue formations, and dangerous driving behaviors. This study rigorously evaluates the effectiveness of various communication strategies in mitigating the impacts of these events on the network. Specifically, a comprehensive simulation framework employing Dynasmart dynamic traffic assignment models is developed to account for mixed traffic flows, considering both human-driven and autonomous vehicles. The evaluation focuses on three aspects: the spatial distribution of information, the temporal reactivity to events, and the nature of the messages conveyed (alerts versus prescriptive instructions). Focusing on a case study in Rome’s EUR district, simulation scenarios assess the optimal dissemination of messages across the most incident-prone hotspots. Findings reveal that real-time, targeted communication markedly diminishes congestion, enhances rerouting efficiency, and bolsters overall traffic safety, positioning IVA as an indispensable tool for intelligent transportation systems.

ABSTRACT. Road traffic accidents present significant public safety and economic challenges, necessitating accurate severity prediction for targeted safety interventions. Traditional models rely on historical crash data but often overlook real-time traffic conditions and road geometry. This study proposes a machine learning framework integrating traffic flow data, road network attributes, and accident records to enhance severity classification. A key focus is assessing the impact of excluding post-accident features to ensure predictions rely solely on pre-crash conditions. A web-based application automates network extraction, traffic assignment, and data integration, making the framework scalable across urban environments. The approach employs XGBoost for severity prediction and SHAP for feature importance analysis, validated using Rome as a case study. Results highlight traffic flow, speed variance, and road design as critical severity determinants.

ABSTRACT. Understanding the effect of road geometry on human driving behavior is essential for road safety studies and traffic microsimulation. Research on this topic is still limited, mainly focusing on isolated vehicles and not adequately considering the influence of curvature on leader-follower dynamics. This work investigates this issue and models the adaptation of leader-follower behaviors to horizontal curvature. For this purpose, the maximum desired speed – which mainly determines the free-flow dynamics – is expressed as a parsimonious function of the curvature. A spatial anticipation mechanism and an indifference threshold are also included to describe realistically the driving behavior when approaching or exiting from curves. The accuracy of the augmented model is evaluated using the Intelligent Driver Model (IDM and M-IDM) and trajectory data from isolated vehicles, free-flow, and car-following traffic (Naples Data and Zen Traffic Data). The results show that an improvement is achieved in isolated vehicle and free-flow dynamics. In car-following situations, the improvements are less pronounced and depend on the observed driver. Overall, the analysis highlights the lack of sufficiently spatially extended trajectory data to calibrate and evaluate such driving behaviors.

ABSTRACT. Heterogeneous driving behavior is often modeled by assuming individual drivers (or agents) have distinct behavior that is time-invariant. However, in reality, individual driver behavior may be influenced by the driving behavior of other drivers in their proximity, and thus change over time. Inspired by recent advances in modeling epidemic spread processes, we introduce the aggressive-passive-aggressive (or APA) traffic model. This model is based on the networked susceptible-infected-susceptible (SIS) epidemic model, where infection level in individual networked populations is modeled based on the level of interaction and level of infection of neighboring nodes in the network. In the APA model, individual drivers are either aggressive or passive, and change their driving behavior based on other drivers around them. By interacting in a local network, drivers change their aggressivity state and thus their driving behavior. We investigate the string stability implications of this model and explore the possibility of controlling the traffic state using a small number of drivers who induce specific behavior into the traffic network. We find that the proposed model is capable of capturing changes in driver behavior that result from driver interactions and may help improve traffic control strategies.

ABSTRACT. To evaluate traffic management policies and new in-vehicle technologies, a digital simulation environment can be useful for testing their functions in virtual experiments. This paper presents a methodology for building a traffic model in a microsimulation environment (SUMO) based on observed vehicular trajectory data. Applying an iterative process, the methodology to refine the traffic model is described to obtain output statistics consistent with observations. In particular, distributions of individual travel times and distances are comparable for most of the simulated and observed vehicles. The main errors detected during the model building phase were fixed, and the solutions adopted were detailed concerning vehicle route selection, traffic light setting, and manoeuvres modelling.

ABSTRACT. This study enhances the calibration of traffic microsimulation models (TMMs) by incorporating lane change (LC) dynamics into the objective function, leveraging re-identified individual vehicle data (RIVD). Focusing on a complex motorway weaving section, the methodology integrates LC probabilities and temporal distributions alongside traditional traffic variables such as speed and headway. Results reveal that including LC data significantly improves calibration efficiency and parameter consistency, particularly for LC-related parameters. Incorporating LC temporal distributions ensures robust alignment with real-world traffic patterns, addressing non-stationary behaviors. This approach demonstrates the importance of detailed LC modeling for accurately representing merging and weaving dynamics, where lateral maneuvering dominates the individual and aggregate traffic dynamics.

ABSTRACT. Evaluating autonomous vehicle performance in complex traffic scenarios requires both calibrated sensors and realistic traffic conditions. However, existing traffic datasets focus on vehicle trajectories but lack comprehensive sensor data, while perception datasets provide sensor data but offer limited traffic diversity. To bridge this gap, we present an open-source toolset that generates perception datasets in CARLA by integrating real-world vehicle trajectories into a simulated sensor framework. This approach creates realistic sensor data mirroring real traffic dynamics, providing a valuable resource for testing AI-driven autonomous systems in challenging scenarios like roundabouts, where sensor-based perception is critical for safe navigation.

ABSTRACT. The technology of autonomous vehicles (AV) is developing rapidly and is expected to bring systematic changes to the way we travel. An increasing number of studies focus on people's preferences, acceptance, and intention to use autonomous vehicles, although they still rarely appear in regular road traffic. Understanding the opinions of users and experts in the broadly defined field of transportation regarding these innovative technologies will enable actions aimed at increasing societal acceptance of automated and autonomous vehicles. This article presents selected research findings conducted on a representative sample of the Polish society as well as among experts. The level of acceptance does not significantly differ from that identified in studies carried out in other countries. Increasing this acceptance requires actions of various types, including organizational, legislative, technological, and educational measures. The article also presents expert-formulated suggestions for actions aimed at enhancing the efficiency of autonomous vehicle implementation in the economy.

ABSTRACT. The development of Advanced Driver Assistance Systems (ADAS) requires a controlled testing environment for evaluating perception, decision-making, and control algorithms. This paper presents a connected infrastructure for ADAS testing, integrating roadside LiDAR and stereo cameras, a 5G-based communication network, and an instrumented prototype vehicle. The development included both a digital twin for virtual testing and a physically instrumented real-world area. The system supports key scenarios, including automated parking, safety-focused ADAS for Vulnerable Road Users (VRUs), and collaborative localization. A case study on LiDAR-based cooperative localization demonstrates improved positioning accuracy, highlighting the benefits of the connected infrastructure for autonomous driving and the potential of a custom-designed test area for this type of applications.

ABSTRACT. Traffic congestion at motorway on-ramps is a persistent bottleneck in urban networks. This study investigates a Cooperative, Connected, and Automated Mobility (CCAM) strategy for mitigating such merge congestion, focusing on a representative cooperative merging service known as Gap-Opening. In a case study on the A56 — Tangenziale di Napoli — in Italy, we employ a high-fidelity co-simulation platform (Eclipse MOSAIC) to evaluate four scenarios: a no-intervention baseline, a conventional temporary lane-closure control, and CCAM deployments with 40% and 50% connected and automated vehicle (CAV) penetration. The Gap-Opening service enables mainline CAVs to proactively create gaps for merging vehicles based on real-time Vehicle-to-Everything (V2X) communication from on-ramp vehicles. Simulation results show that the cooperative strategy significantly outperforms the traditional approach, yielding shorter average travel times and more stable traffic flow. These findings demonstrate that integrating CCAM merging services can be a more effective alternative to reactive measures, supporting smoother merging operations and improved traffic performance in congested motorway scenarios.

ABSTRACT. This paper proposes a parallel simulation framework for the advanced testing of connected and automated vehicle (CAV) systems. The platform integrates an orchestrator framework, open-source network and traffic simulator, and industrial Hardware-in-the-Loop system, thereby providing a synchronised environment for the study of traffic behaviours, communication delays, and vehicle dynamics. The integration of message broker services and real V2X communication processes facilitates capturing message timing and network-induced delays. The incorporation of accurate environmental modelling, including physical obstacle placement, ensures realistic representation of propagation and shadowing effects. This simulation environment enables comprehensive performance evaluation of CAV applications under various traffic and network conditions, offering insights into latency-critical scenarios. The proposed approach addresses the disparity between theoretical analysis and practical implementation, thus facilitating the development of resilient and efficient vehicle technologies through the provision of precise latency and performance measurements.

ABSTRACT. To reduce urban congestion and use existing road space more efficiently, it will be crucial in the coming decades to increase vehicle occupancy and encourage ride-sharing. This can be achieved by prioritizing high occupancy vehicles (HOVs) at intersection control or by measures such as an HOV lane. In a future with 100% connected and automated vehicles (CAVs), Autonomous Intersection Management (AIM) will be the most effective intersection control, controlling vehicles without traffic signals and significantly reducing vehicle delay compared to today’s traffic signal control (TSC), even when pedestrians are involved. This paper is the first to implement person-based AIM, which considers individual vehicle occupancy and waiting pedestrians in optimization and examines various control schemes to prioritize HOVs and incentivize ride-sharing. We simulate person-based AIM under different control schemes for a real intersection using the microscopic simulation platform SUMO and benchmarking it against vehicle-based AIM and today’s traffic-actuated signal control (TSC). CAVs in our study have 0 to 4 passengers, with HOVs defined as having 3+ passengers. The results show that with person-based AIM it is possible to achieve shorter travel times for HOVs. Elevating HOV delay in optimization is an additional strategy that benefits HOVs e.g., with an average occupancy of 2.0. Combining person-based AIM with an HOV lane is by far the most effective strategy at low average occupancy rates and can reduce HOV travel times by up to 70% compared to those of low-occupancy vehicles (LOVs).

ABSTRACT. On-demand pooled-ride services represent a possible method of mitigating congestion delays in urban areas. However, as service penetration increases they may have an impact on both link and network performance. This paper proposes a framework for solving the dynamic dial-a-ride problem considering the mutual interactions between private vehicle and service fleet routing decisions. To capture network traffic evolution a multi-class dynamic system optimal traffic assignment problem is used to model private vehicles' optimal route choice considering planned fleet routes. Traffic state evolution is considered in the dial-a-ride problem through updated link travel times. Numerical examples demonstrate the importance of considering mutual traffic interactions for both network state estimation and on-demand service performance.

ABSTRACT. Integrating demand-responsive mobility services with transit systems is recognized as a practical and effective strategy to mitigate their impact on traffic congestion and the environment. This paper develops an efficient hybrid metaheuristic to solve the corresponding integrated dial-a-ride problem utilizing electric vehicles to minimize both operational costs and customer travel time. The system aims to improve customer convenience by limiting a maximum intermodal transfer time to synchronize demand-responsive buses’ arrival and transit departures. The proposed metaheuristic addresses the challenges of optimizing the integrated demand-responsive vehicle routing and charging operations with fixed-route transit systems with capacitated charging stations and partial recharge. The algorithm is tested on instances with up to 50 customers, outperforming an 8-hour state-of-the-art solver by 23% in solution quality, with an average runtime of 136 seconds.

ABSTRACT. This study investigates the integration of Vehicle-to-Grid (V2G) technology within ride-hailing fleets, focusing on the implications for profitability and operational efficiency. Through the adoption of agent-based mobility simulation paradigms and the development of a novel two-sided mobility platform simulation framework considering electricity grid dynamics, this research evaluates various charging strategies, including traditional models, smart charging and V2G across different electricity price schedules, fleet sizes and demand configurations. Findings indicate that V2G outperforms traditional models, achieving an average increase of 18\% in daily fleet profits, while maintaining low trip rejection rates. Nonetheless, V2G integration introduces additional operational complexities, including increased travel distances and more frequent charging and discharging events. These factors can lead to negative externalities (e.g., congestion) and battery wear.

ABSTRACT. An origin-destination (OD) matrix is a fundamental tool in urban transportation planning, representing the movement of people or goods within different zones of a city. In this paper, a novel approach based on ensemble learning is proposed to address the challenging task of estimating mobility demand based on volunteered geographic information (VGI) and population statistics. A hierarchical estimation strategy is adopted. Given an OD zoning system, first, for each zone, the relationship between input VGI and population statistics data, and the total mobility demand generated and attracted by the zone is modeled. Then, the OD matrix entries are estimated, by conditioning on the resulting total demands. Each estimation step is formalized as a supervised regression addressed using an ensemble of regression trees (namely Random Forest). The proposed approach is experimentally validated on a case study associated with the metropolitan city of Genoa, Italy, whose territory is composed of 71 traffic zones. The proposed approach achieves root mean square error and R2 values of 438.05 and 62.93%, respectively, which suggests the effectiveness of the proposed approach despite the complexity of the problem of estimating the OD matrix.

ABSTRACT. This study presents a novel integration of Graph Neural Networks (GNNs) with Simultaneous Perturbation Stochastic Approximation (SPSA) to enhance the computational efficiency of Simulation-based Offline Origin-Destination (OD) Matrix Calibration. The proposed framework leverages a GNN-based metamodel as a surrogate for computationally intensive transport simulations, utilizing a Node-to-Node Graph Regression (N2N-GR) model trained on 7200 historical simulation datasets. The metamodel effectively approximates traffic flow predictions from OD matrices, thereby significantly reducing the computational overhead associated with traditional SPSA implementations. The findings demonstrate that GNN-based SPSA calibration can achieve substantial efficiency gains while maintaining solution fidelity, presenting a scalable paradigm for advancing traffic demand modeling and transportation system optimization.

ABSTRACT. Origin-Destination (OD) matrices are essential inputs to both traffic planning and management. To enable traffic management decisions, the OD matrix needs to be estimated in the order of minutes, which is very challenging in many situations. However, new large-scale mobility data, such as vehicle probe data, in combination with computationally efficient estimation methods make it possible to estimate OD matrices sufficiently fast to enable traffic management decisions. In this paper, we propose a computationally efficient method that uses link count data and vehicle probe data, within a data-driven network assignment (DDNA) framework in combination with non-negative matrix factorization (NNMF), to estimate OD demand in urban networks. Historical data are used to construct a low-dimensional description of the OD estimation problem using non-negative matrix factorization. The method and the quality of the OD matrix estimated using the low-dimensional representation are evaluated on empirical data for central Stockholm, Sweden. The results demonstrate that the method provides a computationally fast technique to find an OD matrix that provides link flow estimates close to the link flow observations for both training and test sets.

ABSTRACT. As urban populations increase and traffic congestion escalates, public transportation systems, especially buses, provide a sustainable solution by decreasing the reliance on private cars and minimizing fuel consumption. However, operators must address passengers' concerns about long waiting times and overcrowded conditions to keep buses an attractive option. Therefore, real-time predictions of passenger demand are essential for optimizing scheduling, reducing headways, and enhancing service reliability. Despite its importance, short-term forecasting of bus passenger demand is still underexplored, facing challenges such as seasonal fluctuations, periodicities, and interactions with other transport modes. This paper introduces a new study to predict bus station demand patterns using Google Popular Times (GPT) data through a two-step deep learning approach. Drawing on real-world historical data from bus stations, we propose a predictive framework that starts by classifying passenger demand at each station into distinct clusters. Sequence-to-sequence (Seq2Seq) models are subsequently trained for each cluster to predict demand patterns for the next 24 hours, using the previous 72 hours of data as input.