ABSTRACT. The paper describes a procedure for simulating the recharging activities of private electric cars in urban areas, thus enabling the reconstruction of daily electric power demand profiles on a zonal basis. The procedure relies on a behavioral model, which is calibrated using data from a large Stated Preference (SP) survey. The results of this calibration, along with pre-processed Floating Car (FC) data, serve as inputs for a simulation model, allowing for the integration of both behavioral insights and real-world travel patterns. The procedure compares, with a “what – if” approach, different demand and supply scenarios, varying the distribution and technical characteristics of the electric cars fleet as well as the level of both private and public charging infrastructure in the study area. The goal is to identify charging supply networks that meet demand needs while minimizing installation costs and local power requirements. The procedure has been applied to a sample of approximately 10,000 cars circulating in the metropolitan area of Rome, for which it was possible to identify the area of residence. Future developments could involve re-conducting the survey to refresh model parameters, validating the results with actual charging data, and assessing the potential for smart charging and vehicle-to-grid (V2G) scenarios.

ABSTRACT. The rapid technological evolution in the automotive sector demands increasingly advanced testing platforms capable of integrating high-fidelity virtual simulations with real hardware components. In this work, we present a modular platform capable of operating in Software-in-the-Loop (SIL), Vehicle-in-the-Loop (VIL), and Passenger-in-the-Loop (PIL) configurations. The proposed solution combines photorealistic rendering using Unreal Engine 5, real-time processing via SpeedGoat and MATLAB/Simulink, advanced tracking systems (Vicon, IMU, GNSS RTK), and an Adaptive Traffic Controller for autonomous management of heterogeneous traffic scenarios. The platform is designed to simulate complex urban environments while ensuring low latencies (average below 24 ms) and minimizing motion sickness risk—key aspects for validating autonomous and assisted driving systems. Preliminary results highlight the correct integration of the various modules, demonstrating the system's ability to manage real-time interactions between vehicles, pedestrians, and virtual scenarios. These developments provide a valuable tool for optimizing and validating control algorithms, paving the way for safer, more reliable, and immersive testing.

ABSTRACT. Abstract— This paper introduces a novel hybrid control strategy for autonomous driving that integrates neurotechnology-inspired signal modeling with conventional PID control to enhance vehicle performance and adaptability. The proposed approach combines a standard, fixed-gain PID controller with a dynamic neuro-fuzzy system that adjusts controller gains in real time via a dual-layer fuzzy logic framework. Specifically, a "Fuzzy Brain" module estimates the driver's Emotive State from surrogate inputs—such as lateral acceleration and yaw frequency ratios—while a "Fuzzy PID" module modulates the PID gains accordingly. Vehicle dynamics are modeled using the widely adopted bicycle model, and numerical simulations are conducted in a cone slalom scenario at speeds of 50, 75, and 100 km/h. Results indicate that under high-demand conditions, the hybrid system yields improved trajectory tracking, stability, and responsiveness. Moreover, when the Emotive State is relaxed, the system prioritizes comfort by adopting less aggressive gains, thereby providing a balanced trade-off between agility and ride quality. These findings demonstrate that incorporating cognitively inspired inputs into autonomous control architectures can pave the way for safer, more adaptable driving systems.

ABSTRACT. In this paper, a new transport system for city logistics, based on automated vehicles, is presented. Deliveries are performed by platoons of electric and automated vehicles in which the leading one is driven while the others move driverless. Each platoon travels from the Urban Distribution Centre (UDC) to a specific place on the border of the city centre, named Split Up Location (SUL), where the platoon is divided. Then each vehicle, except for the leading one, carries out the last mile deliveries in a driverless way. Meanwhile, the driver relocates to another SUL or to the UDC, if required. After finishing the deliveries, vehicles come back to the SUL and reassemble in a platoon; a driver gets into the leading vehicle, then the platoon returns to the UDC. A methodology has been developed to design, simulate and optimize the proposed transport system. The proposed methodology is composed of: optimization algorithms for all phases of the delivery trips; an object-oriented simulator of the proposed transport system. The simulator modelizes the second-by-second activities of each platoon, each vehicle and each driver and it allows to evaluate, dynamically, the number of vehicles required to operate the system and the timetable of driver activities. The system proposed has been applied to the case study of Piraeus, Greece.

ABSTRACT. The advancement of autonomous electric vehicles presents new opportunities for sustainable and efficient public transportation. This paper explores an innovative approach that integrates shared dynamic charging, convoy automation, and a novel mechanical coupling mechanism to enhance safety in traveling convoys for urban mobility. The proposed system enables autonomous vehicles to form modular convoys, dynamically adjusting capacity based on passenger demand. An intelligent energy management framework allows for real-time power distribution among vehicles through on-the-go charging, leveraging a single-wire catenary infrastructure. A nonlinear model predictive control (NMPC) strategy ensures precise trajectory tracking and adaptive leader-follower transitions within convoys. Additionally, a fail-safe mechanical coupling mechanism facilitates the delayed transfer of steering, ensuring stability and safety even in the event of electronic failures. Simulation results validate the effectiveness of the control strategy, while ongoing prototype development aims to assess the feasibility of the mechanical and charging solutions. This research contributes to the development of sustainable, flexible, and resilient urban transport networks.

ABSTRACT. RECIFE-MILP tackles the real-time railway traffic management problem. It relies on the solution of a mixed integer linear programming (MILP) model through a MILP solver. Mostly linked to this solver, RECIFE-MILP includes numerous configurable parameters, which can significantly influence its performance and are challenging to configure manually. Consequently, automatic parameter configuration appears as a promising option to obtain the best possible performance in a short computation time. This paper addresses two research questions: (1) Can automatic parameter configuration significantly enhance RECIFE-MILP’s performance? (2) Do distinct classes of problem instances require specialized configurations? To answer these questions, we present an experimental study using irace, a state-of-the-art parameter configuration tool. Experiments on three French control areas demonstrate that automatically chosen configurations achieve significant performance gains compared to default parameter settings. However, the benefit of configuring RECIFE-MILP for specific classes of instances does not clearly emerge.

ABSTRACT. Metro networks face operational challenges due to increasing ridership and system growth, particularly in managing delay propagation. Epidemiology models have recently been an interesting method in transportation research for studying delays. This study, therefore, aims to investigate if the Susceptible-infectious-susceptible (SIS) model is suitable to help model delay propagation in a metro network through its ability to reproduce the vulnerability of metro stations for specific instances. Using data from the Washington Metro Network, two groups of delay propagation instances were selected and used for model training and testing using a differential evolution algorithm. The results indicate that the vulnerability values as calculated from the real life data do not follow the expected trend. Still, our model can capture this variation with good vulnerability estimation accuracy for both groups. Also, the predicted vulnerability values for the first group are more accurate than for the second group. However, limitations such as underestimation and overestimation of station vulnerabilities, and sensitivity to training data were observed. These challenges stemmed from the dynamics between specific parameters and the lack of additional factors.

ABSTRACT. This paper proposes an optimization algorithm to support decision makers in the dimensioning of a new type of public transport service. In particular, vehicles capable of traveling on both rail and road are exploited to revitalize capillary railway lines, characterized by low demand but very high societal impact. To support decisions on infrastructure modifications and fleet size, we propose an algorithm based on the solution of a mixed-integer linear programming model. In particular, the infrastructure modifications considered consist in the construction of road-type sidings on the existing railway line, to allow vehicle crossing. We assess the performance of our algorithm on a real case study: the line joining the cities of Limoux and Quillan, in France. The results highlight how investing in siding construction can indeed improve the regularity of the timetable that can be planned. However, depending on the available fleet and the desired frequency, they may not be strictly necessary.

ABSTRACT. A common approach to mitigate delays in railway networks and improve timetable robustness is to incorporate time supplements. These supplements allow trains to accommodate small disturbances and reduce knock-on delays, often by permitting trains to run at their maximum allowed speed. However, most studies focusing on robustness and time supplement allocation do not account for real-time corrective actions. The concept of fragility offers the potential to improve this decision-making process. In this work, we present a comprehensive analysis of the fragility of train timetables to study its potential to help route planners implement more targeted measures to enhance its robustness. This study investigates the impact of infrastructure constraints, such as station capacity, on the fragility of a train timetable. Preliminary results from real-life scenarios of the Jæren Line in Norway are presented.

ABSTRACT. The European Railway Traffic Management System (ERTMS) aims to replace the heterogeneous railway Traffic Management Systems implemented in the different countries of the European Union, promoting interoperability between national systems and, consequently, improving the efficiency of traffic management and facilitating the transit of international freight and passenger trains.

One of the main challenges of ERTMS is meeting the high safety constraints of the railway domain, which becomes even harder considering the digitalization process of transportation systems and their integration with a wide range of services with different Quality of Service requirements. Particularly, while considering a safety-critical ICT platform, one of the most challenging issues is to guarantee time constraints of time-critical tasks in the worst case scenario. This issue can become even more challenging when trying to increase the set of functionalities the system offers, as happens in the digitalization process.

In this paper, a hybrid-virtualized architecture for ERTMS applications is proposed and preliminary tested whose aim is to operate as an ecosystem where hard real-time tasks can run together with soft real-time tasks offering an applications environment flexible, extensible, and, at the same time, suitable to meet strict temporal and high safety requirements, becoming a suitable candidate for the next generation ERTMS applications.

ABSTRACT. The increasing urbanisation and traffic congestion have driven interest in AI-based traffic signal optimisation, particularly through automated planning techniques that offer transparency and explainability. Existing automated planning models range from highly flexible, that assume minimal infrastructure constraints, to fully deployable ones that can accommodate constraints of existing legacy systems.

In this work, we introduce TRADE, a novel knowledge model for planning-based traffic signal optimisation that bridges the gap between the two classes of models by enforcing practical constraints, such as cycle length limitations and consistency between consecutive cycles, while maintaining the benefits of flexible approaches. Empirical evaluation using real-world data from West Yorkshire, UK, shows that the TRADE model allows to deliver comparable performance to fully flexible models while enhancing deployability, contributing to more practical and effective AI-driven traffic control solutions.

ABSTRACT. The Traffic Assignment Problem is a fundamental, yet computationally expensive, task in transportation modeling, especially for large-scale networks. Traditional methods require iterative simulations to reach equilibrium, making real-time or large-scale scenario analysis challenging. In this paper, we propose a learning-based approach using Message-Passing Neural Networks as a metamodel to approximate the equilibrium flow of the Stochastic User Equilibrium assignment. Our model is designed to mimic the algorithmic structure used in conventional traffic simulators allowing it to better capture the underlying process rather than just the data. We benchmark it against other conventional deep learning techniques and evaluate the model's robustness by testing its ability to predict traffic flows on input data outside the domain on which it was trained. This approach offers a promising solution for accelerating out-of-distribution scenario assessments, reducing computational costs in large-scale transportation planning, and enabling real-time decision-making.

ABSTRACT. The design and development of intelligent transport systems rely heavily on the availability of data, which allows for a more comprehensive understanding of current traffic conditions and their likely evolution. While merging all data sources together might be tempting, ensuring the consistency of data collected from different sources is crucial. Noise or large discrepancies can jeopardise the usefulness of merging data and hinder the potential benefits. This paper investigates the integration of data from different sources by analysing the consistency of data from two sensor sets deployed in a region of Manchester, United Kingdom. To perform this analysis, we identified suitable road segments for consideration and leveraged extended Kalman filters. Further, we exploit the opportunity to assess the sensitivity of the sensors to potentially critical circumstances.

ABSTRACT. The integration of real-time multimodal traffic data fusion is critical for advancing Intelligent Transportation Systems (ITS) in urban environments. This paper presents a novel camera-based traffic data collection system deployed at the Munich Mobility Research Campus (MORE) as a real-world laboratory. The system captures multimodal trajectories in real-time, which are processed using a state-of-the-art data fusion engine, integrating both low- and high-level fusion functionalities. The evaluation focuses on challenging traffic scenarios, including mixed-traffic environments, groups of vulnerable road users, and adjacent cycling pathways. Performance assessment for aggregated traffic counts is conducted using Root Mean Square Error and paired t-tests, while isolated detection is evaluated using True Positive Rate, Positive Predictive Value, and F1-Score. Results indicate high accuracy in detecting and classifying vehicles and bicycles, with minor underestimation biases observed in pedestrian detection within dense groups. The findings demonstrate the potential of real-time trajectory data fusion for multimodal ITS applications, supporting enhanced traffic planning, management, and control. Future work will extend this research by analyzing system latencies, refining speed accuracy, and exploring real-time anomaly detection. The proposed system contributes to the deployment and evaluation of scalable ITS solutions for connected, automated, and multimodal mobility.

ABSTRACT. In Belgium, floods are acknowledged as one of the most frequent natural disasters, posing serious threats to people's lives and property. Growing evidence suggests flood risks will intensify in the coming decades, driven by climate change, population growth, and evolving land use patterns at the catchment scale. These compounding factors make improved flood risk understanding and management an urgent priority. The impact of floods on the transportation system primarily stems from road interruptions, which significantly affect travel demand. Exploiting mobile phone data collected by providers makes it possible to geolocate mobile phone users over time to derive time-dependent crowding maps. By intersecting these maps with flood inundation data, we can quantify human exposure to flood risks. This integrated approach enables a detailed analysis of both the spatial extent of floods and temporal changes in population movement patterns during flood events. In this context, we propose a sensitivity analysis based on mobile phone data collected from pre- and post-flood calendar periods in the Vesdre catchment area (Wallonia, Belgium). In light of the floods that occurred in July 2021, mobile phone data collected in 2018 and 2022 have been processed and compared. Meanwhile, we investigate the impact of the transportation infrastructure disruptions on mobility within the Vesdre catchment area and apply a Tobit regression model to analyze the significant parameters. As a result, we observe a decrease in interaction between the valley and the plateau, except between urban centers in the valley and neighboring residential communes in the heights, along with a general decline in mobility in the most affected communes. This suggests that the flooding has incited people to get further away from the river. Besides, we find that the parameter representing the number of out-of-service transportation infrastructures per kilometer is significant in the 2022 flow regression.

ABSTRACT. Understanding commuter behavior is essential for developing effective and sustainable urban transportation systems. As travel patterns evolve due to infrastructure changes, emerging mobility services, and shifting commuter preferences, accurately predicting travel choices becomes increasingly challenging. This study evaluates different modeling approaches to enhance the reliability and adaptability of predictive frameworks across various urban and behavioral contexts. By leveraging real-world commuter data from two major Italian cities, we systematically assess the strengths and limitations of different predictive strategies in capturing behavioral dynamics and their generalizability across diverse mobility environments. Using a dataset from Rome (2022 and 2023) and Turin (2023), we conduct a 10×5-fold cross-validation to assess in-sample predictive accuracy and apply ANOVA statistical tests to quantify performance differences. Crucially, we extend traditional validation methods by evaluating model transferability, testing the ability of models trained in one city to predict commuting behavior in another, as well as within the same city across different time periods, particularly in the post-COVID context. Our results reveal that while some approaches achieve higher predictive accuracy, others demonstrate greater robustness in gen- eralization across cities and temporal shifts. These findings highlight the trade-offs between predictive performance, interpretability, and transferability, offering key insights for urban planners and policymakers aiming to implement data-driven transportation solutions.

ABSTRACT. We address the significant opportunities and inherent challenges in developing advanced mobility observatories, critical tools for managing the profound transformation of urban mobility underway, driven by data proliferation, advances in AI and digital twin technologies. To inform this discussion, we first critically review the landscape of data collection methods – from traditional sources such as travel surveys and traffic counters to emerging streams such as mobile phone and social media data – and highlight the benefits and limitations of each approach. Existing mobility dashboards and observatories are examined to understand their current utility and limitations. Building on this analysis, we present a dynamic observatory architecture proposed for Luxembourg that uses automated Extract, Load, Transform (ELT) pipelines and integrates various open data sources. This experience highlights significant data quality challenges and necessitates mitigation strategies, which are discussed. Crucially, our proposed architecture and the Luxembourg case study highlight the essential role and need for the development of interoperable Local Digital Twins (LDTs). We conclude by advocating that to realise the full potential of next-generation mobility observatories, integrated data spaces and sophisticated AI-driven tools must be adopted for future urban mobility management.

ABSTRACT. Loop-detectors count amongst the most adopted road sensing infrastructure. An emerging use-case for this type of sensor data are digital twin models of cities and highways, that allow for real-time, simulation-based support in safety-critical decision-making of intelligent transportation systems. While previous work used aggregated, counting data from loop-detectors for simulation calibration, the use of non-aggregated, timestamped, event data was overlooked. This study proposes a method to reconstruct single trips from traces that vehicles leave passing a sensor and road network, which goes far beyond mere estimate probabilistic models of origins and destinations. A case study of the arterial network Schorndorfer Strasse in Esslingen am Neckar (Germany) with 34 considered loop detectors and 96 traffic lights, demonstrates the feasibility of the approach. An extensive benchmark with other methods shows, that simulations generated by the method achieve similar levels of accuracy on a macroscopic and mesoscopic level, while significantly improving accuracy on a microscopic level of up to 40%.

ABSTRACT. This article describes a method of the optimal cluster quantity selection the Traffic Analysis Zones (TAZ). It includes macroscopic model trip motivation data selection, clustering method, similarity analysis and cluster optimisation method. The paper shows on the example of the GZM metropolis in Poland the whole process from data extraction to clustering.

ABSTRACT. Potholes pose a major threat to road safety, contributing to traffic accidents and vehicle damage. Early and efficient detection is crucial for mitigating these risks. Traditional detection methods, such as, manual inspections and specialized instrumented vehicles, are costly and impractical for large-scale deployment. In contrast, smartphones, equipped with built-in sensors, provide a scalable and cost-effective solution for opportunistic sensing of road conditions. However, existing smartphone-based detection methods have limitations. Classical machine learning requires manual feature selection, which is time-consuming, while LSTM-based methods suffer from slower processing due to sequential computation. This paper introduces Temporal Convolutional Network with Self-Attention (TCN-SA), a deep learning model designed for accurate and efficient pothole detection using smartphone sensor data. TCN-SA reduces computational costs by leveraging parallel data processing. To improve robustness, we collected real-world data using a smartphone mounted on a stand and applied a high-pass filter to reduce noise. Experimental results demonstrate that TCN-SA achieves 90.4% accuracy while reducing inference time by 25% compared to LSTM-based model, making it ideal for real-time pothole detection. The code of this work is made publicly available at: https://github.com/ZJU-TSELab/pothole-research.

ABSTRACT. Mobility as a Service (MaaS) is reshaping urban transportation by integrating multiple mobility options into seamless digital platforms. Despite its transformative potential, uncertainties persist regarding its scalability and economic viability. This study bridges theory and practice by developing MaaS business model archetypes that guide providers and policymakers in structuring effective services. Using a multiple-case study approach, data from 40 MaaS companies were analyzed through the Business Model Canvas framework. Five distinct archetypes emerged: Mobility Platforms, Highly-diversified Providers, Lowly-diversified Providers, Enablers, and Mobility Embedded Providers. Each archetype reflects different strategic configurations, value propositions, and revenue mechanisms. The findings offer a structured understanding of MaaS business models, supporting industry stakeholders in refining their strategies and informing policymakers on regulatory frameworks that foster innovation and sustainable mobility. By linking conceptual insights with real-world applications, this research provides practical guidance for advancing integrated urban mobility solutions.

ABSTRACT. Mobility as a Service (MaaS) has the potential to transform urban transportation by integrating multiple mobility services into a seamless and user-centric system. However, the impact of different MaaS subscription models on transport demand and system efficiency remains an open challenge. In this research, we propose a simulation-based methodological framework to evaluate the effects of MaaS implementation on users’ modal choices, operational performance, and sustainability. The framework integrates discrete choice models for demand estimation and macro-simulation tools for network representation, allowing for the computation of quantities of interest for multiple stakeholders of the MaaS ecosystem. A synthetic urban network is used to test various MaaS configurations, analyzing key performance indicators such as modal shift, and revenue distribution among operators. Results indicate that different MaaS bundles generate distinct effects on travel behavior and system performance, highlighting the need for a structured assessment before implementation.

ABSTRACT. In the modern era of transportation in urban environment, optimization of traffic is crucial. In the context of Smart Cities, Mobility as a Service (MaaS) is a revolutionary paradigm that offers integrated solutions to enhance urban mobility. The general goal is to develop a single digital hub that allows users to plan, book, and pay for their journeys, as well as gather data about public transport, their occupation, traffic, and other alternative modes of transportation (like e-mobility solutions, car sharing, bike sharing, etc.). In this context, security and privacy of data and infrastructures, are still open issues that may limit full adoption of these applications. In this paper, we propose a reference architecture to improve security in MaaS platforms. We present the initial design process to understand the requirements needed to cover all the characteristics of a MaaS system that defines a reference architecture; later, security analysis will be performed with a risk-based and threats-based approach. Finally, as a Proof of Concept, we will present an implementation of a case study to validate the security of the proposed solution suitable for a secure distributed MaaS application.

ABSTRACT. This study investigates a user-centered microtransit service specifically for transporting adolescents to after-school activities. The methodology includes generating a synthetic student population and the corresponding trip requests and solving a Vehicle Routing Problem with Pickup and Delivery with Time Windows (VRPPDTW) for route generation, considering students ’schedules and drivers’ mandatory breaks. The case study focuses on the WeeDrive project, which will introduce an on-demand transport service in Limassol (Cyprus) to help families manage their teenagers’ after-school schedules. This service will operate with pre-booking and a dedicated fleet of private minibuses. Various scenarios are analyzed to assess the impact of different service strategies on the number of students served while ensuring that all trip requests are accepted, as well as the economic viability of the service with a particular focus on user satisfaction.

ABSTRACT. Mobility-as-a-Service (MaaS) represents a different view on how individuals approach transportation, allowing them to integrate various travel modes into a unique subscription, often referred to as a MaaS package/ bundle. This study aims to present an innovative tool designed to collect dynamic stated preferences (SP) data in the context of MaaS. The tool has been designed to collect individual data of respondents, allowing them to customize their MaaS mobility bundles and dynamically reorganize their trip chains based on the selected transport modes. The tool will serve as a basis to collect data with the purpose of training a hybrid Recommendation System architecture, with a double purpose: a) suggesting personalized MaaS packages and b) optimizing real-time single trip choices based on subscribed mobility packages. A case study based on the urban context of Naples (Italy) is presented, highlighting the app's unique components, mainly linked to the transport supply characteristics of the context. The successive analysis of the data will provide valuable insights on user behavior preferences, willingness to adopt and pay for MaaS services, as well as the possibility to perform system-level evaluation based on the collected micro-data.

ABSTRACT. The paper presents the ongoing research and development effort aimed at developing an integrated methodology and an interoperable suite of tools for walkability and bikeability assessments, targeting both the demand for research aides as well as for planning and design support. The suite integrates assessment protocols and software tools across the pipeline from on-site audit protocols for comprehensive micro-scale assessment of urban fabrics, to large-scale massive evaluation of urban areas using big data and machine learning techniques, including research protocols for empirical calibration, validation and targeting using choice modelling techniques and immersive scenarios.

ABSTRACT. Micro mobility operators face a significant challenge in optimally determining their daily inventory levels. An analysis of data from station-based and free-floating services in Boston (USA) and Paris (France) reveals that optimal inventory levels are inherently imbalanced, as the majority of stations require no operator intervention, while a minority necessitate substantial adjustments. From a purely data perspective, the fact that some observations are more frequent than others is referred to as Data Imbalance, and it is a well-known problem when working with Machine Learning models. Therefore, this research proposes testing the Balance Cascade (BC), an algorithm designed to handle data imbalance, to solve the inventory problem. It also proposes an enhanced version of the model able to handle multiple outputs, and a new way to compute the penalty factor. The effectiveness of the approach is evaluated using data from Boston's BlueBike station-based bike-sharing system, demonstrating that BC significantly improves prediction accuracy, and that the proposed enhanved version further enhance performances thanks to a more realistic understanding of the penalty factors.

ABSTRACT. Connected and Autonomous Vehicles (CAVs) and micromobility devices are constant consumers and producers of trajectory data for different safety applications in Intelligent Transportation Systems (ITS). Trusting the truthfulness of a node trajectory is feasible using blockchain based validation via consensus among nearby nodes leveraging wireless communication. However, there is limited work on efficiently detecting malicious nodes generating partially or completely false trajectories in real-time. To address this important problem, we propose a novel transformer based approach to mitigate such data poisoning attacks and detect anomalous trajectory data shared by CAVs and other devices in real-time. Nodes employ a transformer model as an anomaly detection algorithm along federated Q learning based adaptive update of nearby nodes trajectory truthfulness on blockchain under varying network conditions, following a consensus process. The proposed blockchain based data poisoning detection and mitigation scheme is evaluated using trajectory data from New York City, where show its robustness in avoiding malicious nodes to generate false trajectories in a network size of 150 nodes simultaneously validating each others trajectories in real-time.

ABSTRACT. Traffic safety in automated vehicle (AV) research focuses on ensuring safe interactions with other road users. The key challenge lies in understanding and predicting vulnerable road users (VRUs) behavior in traffic, with cyclist-related research remaining relatively limited compared to pedestrians. To address this research gap, this paper proposes a two-stage deep learning model for predicting cyclists’ maneuvers at an unsignalized intersection. The model is based on Bidirectional Long Short-Term Memory network (B-LSTM) and predicts tactical maneuvers (left turn, right turn, or straight crossing) before cyclists enter the intersection. Data is collected using a bicycle simulator that enables interaction with simulated road users in a virtual reality (VR) environment. The cyclist video data serves as the only input to the model, eliminating the need for external devices. The first stage consists of two parallel models that classify cyclists’ explicit and implicit gestures, which serve as communication signals. In the second stage, these classified gestures are used to predict future tactical maneuvers, with predefined gesture weightings assigned based on their correlation with maneuver categories. Results demonstrate the importance of incorporating cyclists’ communication signals, especially implicit gestures, and predefined gesture weightings over model development in prediction accuracy and robustness.

ABSTRACT. Adopting electric scooters in urban environments raises safety concerns, particularly at intersections where interactions with other road users are complex. Geofencing, a technology that enforces automated speed limits in designated areas, presents a potential solution to enhance rider safety. However, its effects on rider behavior and trust remain unclear. This study (n=19) examines the impact of geofencing-based speed limitations using a virtual reality (VR) simulation integrated with a 3-Degree-of-Freedom (3DoF) motion platform. Nineteen participants navigated controlled urban intersections under both geofenced and non-geofenced conditions. Results indicate that geofencing influences head movement behavior, with increased situational awareness by a wide scanning area in some scenarios but more focused scanning area in others. While some riders perceived geofencing as beneficial for safety, others reported decreased trust and increased distraction. These findings highlight the need for user-centered geofencing implementations that balance safety, trust, and usability in micro-mobility. Future research should explore adaptive geofencing strategies and real-world applications concerning infrastructure optimizing its effectiveness.

ABSTRACT. This paper examines the integration of Connected, Cooperative, and Automated Mobility (CCAM) in public transportation, focusing on automated buses. It identifies key challenges such as legislative frameworks, human factors, and infrastructural conditions that impede deployment in Europe. The paper proposes a combination of standardized processes and interfaces, scalable software architecture as well as V2X communication to address these issues. In addition, it discusses projects like MINGA, OTRACE, and Country2City-Bridge (C2C) that exemplify the application of these technologies. Lastly, the gradual introduction of autonomous transport technologies, such as platooning, is recommended to evaluate and test the automated capabilities as well as build public confidence before full automation.

ABSTRACT. Recent advances in battery technologies and a global push for greener transport have accelerated the development of electrified public transportation systems. Such systems often face challenges due to the need for large battery capacities and the high costs associated with conventional charging infrastructure. This study examines the potential of Mobile Autonomous Charging Pods (MAPs), which are autonomous charging vehicles, as an innovative solution to enhance both the efficiency and cost-effectiveness of electric bus operations in urban environments. Using the case of inner-city trunk bus lines in Stockholm and employing a microscopic simulation-based study, three charging scenarios are evaluated: depot charging only, depot combined with end-station charging, and depot plus MAP charging. The results indicate that the integration of MAPs can significantly reduce the required battery capacities and associated infrastructure costs while enhancing the reliability of the service. By facilitating dynamic, on-the-go charging, MAPs offer a sustainable and economically viable alternative for urban electric bus networks.

ABSTRACT. Electrifying demand-responsive transport systems need to plan the charging infrastructure carefully, considering the trade-offs of charging efficiency and charging infrastructure costs. Earlier studies assume a fully electrified fleet and overlook the planning issue in the transition period. This study addresses the joint fleet size and charging infrastructure planning for a demand-responsive feeder service under stochastic demand, given a user-defined targeted CO2 emission reduction policy. We propose a bi-level optimization model where the upper-level determines charging station configuration given stochastic demand patterns, whereas the lower-level solves a mixed fleet dial-a-ride routing problem under the CO2 emission and capacitated charging station constraints. An efficient deterministic annealing metaheuristic is proposed to solve the CO2-constrained mixed fleet routing problem. The performance of the algorithm is validated by a series of numerical test instances with up to 500 requests. We apply the model for a real-world case study in Bettembourg, Luxembourg, with different demand and customized CO reduction targets. The results show that the proposed method provides a flexible tool for joint charging infrastructure and fleet size planning under different levels of demand and CO2 emission reduction targets.

ABSTRACT. The transition to sustainable and climate-friendly transport is a key priority in achieving carbon neutrality. With urban mobility being a major contributor to greenhouse gas emissions, the adoption of electrified public transport solutions is essential. This paper explores the role of intelligent transport systems (ITS) in enabling the deployment and operation of electric bus (eBus) fleets, focusing on both technological advancements and operational strategies. It examines the expansion of electric buses, analyzes various charging strategies and investigates how Internet of Things (IoT)-enabled predictive maintenance can improve fleet reliability and operational efficiency. As a real-world application, the paper highlights the eBRT2030 demonstration in Athens, a pilot project that integrates hybrid charging infrastructure and IoT-based fleet monitoring to optimize the deployment of zero-emission public transport systems. The findings offer practical insights for cities seeking to modernize their transit systems through integrated ITS solutions.

ABSTRACT. The paper aims to analyze the potential effects of truck platooning in terms of fuel consumption and CO2 emissions based on real demand data from the Italian highway system. The approach followed considers only vehicles that share the same entry and exit toll booths as platoonable, an assumption that, while ensuring manageable computational complexity, appears rather limiting for level 1 truck platooning. However, this approach seems highly plausible for level 2 and offers interesting policy insights.

ABSTRACT. This article examines crash statistics and provides a comprehensive analysis of the types, causes, and consequences of road accidents in Japan, focusing on accidents between two vehicles in the central Kansai region (Kyoto-Osaka metropolitan area) with a focus on accidents involving trucks. The data covered the years 2019 to 2022 with 1,296,450 traffic accident records for the whole of Japan and 185,980 in the central Kansai region, which was enriched with measured road data such as traffic volumes. A first analysis with data from the whole of Japan examined the correlation between the collision direction and the fatality. It showed significant differences, with frontal collisions being the most fatal ones (0.77%), followed by side-end (0.26%) and rear-end (0.03%) collisions. Then, two multinominal logistic regression models (MNL, p<0.001) were built, one for studying collisions on road segments and the other for collisions at intersections with the collision direction as outcome variable. These models identified several road features such as guardrails, traffic lights or speed limits which reduce the likelihood of frontal collisions and their installation can therefore be recommended. We were also able to show that several critical factors are related to the drivers. In a final step, based on an analysis of truck route choice preferences, we then aim to bring the results the results together to derive implications as to how road infrastructure improvements would not only directly alter accident likelihoods but also have indirect positive and negative impacts due to resulting changes in routes taken by large vehicles.

ABSTRACT. Truck-drone combined logistics has received increasing attention, and the multi-drops flying sidekick traveling salesman problem (mFSTSP) is one of the classic models. For the mFSTSP, we establishes an integer programming model based on drone range constraint, and designs a sub-circuit construction algorithm according to the modeling idea. Based on the range of the drone, the obtained TSP route is segmented and the optimal sub-circuit are constructed in sequence. Further optimization through dynamic adjustment augmented service mechanism, and three discussions are conducted on the starting position of the segmentation. Considering different operational scenarios, a set of numerical experiments are conducted. Compared with several advanced heuristic algorithms, the proposed sub-circuit construction algorithm shows remarkable advantages in solving large-scale problems, with a maximum solution scale exceeding 700 nodes.

ABSTRACT. The growing demand for efficient and sustainable last-mile delivery has led to the rise of advanced delivery systems, where autonomous delivery robots work alongside vans to enhance logistics efficiency. However, existing models follow rigid routing structures, which makes it difficult to handle parcels split into multiple modules. This paper introduces the Two-Echelon Vehicle Routing Problem (2E-VRP), a novel approach that considers modular parcel logistics within the Physical Internet (PI) paradigm to create a fully connected, traceable, and reconfigurable delivery system. The proposed model ensures that the components of a parcel are consolidated at the final destination while dynamically optimizing transshipment between vans and robots at the robots' drop-off points. By leveraging the enhanced connectivity and real-time data sharing offered by the PI framework, the model optimizes robot capacity utilization and significantly reduces sidewalk congestion. The system minimizes the number of active vehicles, delivery times, energy consumption, and the gap between the first and last module deliveries, thereby improving overall delivery precision. Furthermore, computational experiments comparing the cost functions of the modular and unique parcel formulations have been analyzed and discussed.

ABSTRACT. Detecting traffic anomalies, such as sensor malfunctions and traffic incidents, is crucial to ensuring data accuracy and reliability. However, identifying anomalies in a large urban network is challenging due to the lack of ground truth and the complex spatiotemporal characteristics of traffic data. Traditional methods struggle to differentiate between normal fluctuations and true anomalies. To effectively capture spatial and temporal dependencies, we propose a hybrid deep learning-based autoencoder model, GLA-AE, which integrates Graph Convolutional Networks (GCN), Long Short-Term Memory (LSTM), and a Self-Attention mechanism. To deal with the absence of ground truth, we introduce a data-driven artificial anomaly generation method for evaluation. Our model employs a local thresholding approach for each sensor, ensuring adaptive and robust anomaly detection across diverse traffic patterns. We evaluate GLA-AE on the VAMOS dataset, a large-scale traffic dataset collected from Dresden, Germany. Experimental results demonstrate the model's ability to distinguish between normal traffic variations and true anomalies, and our method outperforms all the baselines.

ABSTRACT. Accurate network-wide traffic volume estimation is essential for optimizing urban mobility, improving traffic management, and supporting environmental monitoring. While extensive research exists on traffic forecasting at sensor-equipped locations, spatial traffic volume estimation—predicting traffic flow at unmonitored sites—remains underexplored. Traditional methods to solve this problem tend to either neglect spatial correlations or rely on complex graph neural networks that are difficult to train and present limited interpretability. To address these limitations, we propose a novel deep-learning approach leveraging attention mechanisms to capture spatial and temporal correlations in traffic data. Our method estimates 24-hour flow profiles for each road segment in a road network using structural network data and historical speed profiles. Results show that the proposed architecture enhances model generalization also to networks whose traffic data were not used for model training.

ABSTRACT. Limited model actionability has recently emerged as a serious concern regarding Deep Learning approaches for traffic forecasting. In this work, we present an efficient multitask modeling framework for predicting the travel times at 30 critical routes of the road network of Athens (Greece) using a single modeling structure fed with traffic flow data from the entire road network. The proposed multitask approach is based on an autoencoder for representation learning and filtering out noisy information. Results indicate that the proposed approach achieves a very reliable performance with a mean prediction error of 11.4%, outperforming all baseline models, even 30 single-task models that were developed to predict the travel times of each route. The proposed multitask model is also significantly more efficient in terms of training time and computational complexity compared to its single-task counterparts, enhancing the model’s actionability. Moreover, SHAP values computed over the predictions issued by the proposed multitask model indicate that there are equally strong relationships between the travel times and the traffic conditions at near and distant locations and that these relationships can be considered almost linear.

ABSTRACT. Urban mobility is experiencing a major shift with the rising popularity of electric vehicles (EVs). To meet the growing demand for EVs, charging stations must offer sufficient coverage, therefore analyzing the current infrastructure is key to be able to forecast future demand. This study presents an innovative data-driven deep learning framework to predict charging demand, integrating data from public parking lot occupancy, charging station usage, and crowdsourced point of interest (POI) popularity. Our paper provides two significant contributions: first, we study the correlation between three data types - charging and parking occupancy, and crowdsourced data. Secondly, we propose a predictive model for EV charging demand integrating parking occupancy and crowdsourced data, disconnecting the prediction from the historical data of the selected charging stations. Consequently, this model is well-suited for application where there are areas lacking charging infrastructure, with the objective of installing new charging stations without the prior knowledge of the demand. Sequence-to-Sequence Recurrent Neural Networks (seq2seq RNNs) were applied with various learning time windows to predict the next 24-hour charging usage pattern. The model, trained on a zone-level and used a 72-hour data window, demonstrated better performance compared to other models, achieving a Root Mean Squared Error (RMSE) of 6.75, a Mean Absolute Error (MAE) of 5.02, and an R² score of 0.80 on the test data.