ABSTRACT. For me, being a Civil Servant in a technical role has been immensely rewarding.  I can look back on my career and see how the evidence I was involved in discovering making it’s way into many of the decisions that have changed the Scottish transport system in the last 20 years (and even further if including my time as a consultant).

It is however, at times, frustrating.  There are only so many hours in the day.  There is only so much budget.  I am responsible for the models, methods and data we use to make strategic decisions.  This is an increasingly fast-paced space which often doesn’t allow the time to commission research.  This means relying on your work to provide the answers needed but sometimes it just isn’t there (or at least discoverable). 

ABSTRACT. The rapid rise of drone technology opens up a wealth of applications, from logistics and disaster relief to personal mobility. Yet, operating large numbers of autonomous drones in unstructured airspaces poses fundamental challenges. This keynote presents a decentralized vision based on differential game theory, inspired by analogies with pedestrian and traffic flow dynamics. We show how game-theoretical strategies enable multi-drone conflict resolution in real time, leading to emergent self-organized patterns such as lanes, diagonal stripes, and roundabouts. These collective phenomena highlight both the potential and the limits of self-organization, motivating the development of macroscopic traffic flow models and the integration of control and regulation when autonomous coordination fails.

ABSTRACT. The Shapley value is a recognized method for allocating total routing costs to individual customers. However, due to its extensive computational intensity, it can only be computed exactly for routing problems with a small number of customers. We propose a new approximation method called β Subcoalition Shapley Approximation. This is based on the mean weighted Shapley values of subcoalitions with exactly β customers. We test this approximation method for the traveling salesman problem as well as for the capacitated vehicle routing problem. Within an extensive numerical study, we demonstrate excellent approximation results representing a good trade-off between approximation quality and computational intensity compared to other approximation methods. Further, our approximation method is not limited to routing problems but can also be applied to other Shapley value applications, especially suitable for NP-hard underlying problems, where the computation time increases dramatically with increasing instance size.

ABSTRACT. Machine learning techniques are increasingly drawing interest in the discrete choice modelling field. There are many potential applications for these. However, the metrics that are used to evaluate models vary across fields and thus far have been overlooked. Specifically, machine learning metrics provide different information compared to the traditional log-likelihood, e.g., focusing on positive or negative predictions of the model. We introduce these measures to choice modelling, estimating these metrics on some classic choice models, and analyse their performance, both overall and disaggregated by mode. The results give more nuanced interpretations of what drives differences in model performance.

ABSTRACT. Aiming to overcome challenges associated with urban mobility, cities adopt different strategies to attain safer and more efficient traffic, among other objectives. An example of such endeavours involves the deployment of Cooperative Intelligent Transport Systems (C-ITS). However, while implementing C-ITS, cities face strong uncertainty regarding different deployment-related aspects, such as the choice of technology and the ideal sequence of services and use cases. This paper, based on data collected by the C-Roads initiative (EU-funded project), proposes a discrete choice modelling approach to predict the ideal implementation period for a specific C-ITS use case. The model aims to assist city authorities in deciding when to implement it, thereby reducing uncertainties in C-ITS deployment. Using data from 22 European cities, the study considers key features that may influence the use case deployment periods, including infrastructure dimension, network design, public and private participation in the several deployment stages, funding sources, and policy goals. The study revealed significant associations between the implementation periods and critical characteristics, such as public-private participation in the deployment stage, the presence of national funding, and environmental concerns being an essential policy goal. Despite limitations arising from data constraints and the complexity of urban C-ITS environments, the model contributes to the decision-making process for a specific use case.

ABSTRACT. This study introduces a hybrid framework that integrates Latent Class Choice Models (LCCMs) with Reinforcement Learning (RL) choice dynamics. Such model allows for: 1) unobserved heterogeneity representation as individuals would be assigned to latent classes with different learning dynamics (e.g., risk averse vs. risk-seeking); 2) class-specific dynamic learning since within each latent class, the choice model would be an RL-based decision process; and 3) explicit modelling of exploration-exploitation trade-off behaviour through class-specific RL models (i.e., individuals may explore new alternatives to gain more information, or exploit a familiar alternative based on accumulated experience).

ABSTRACT. To comply with defined ambitious climate goals, technical and operational improvements are required to reduce the climate impact of aviation. However, a trade-off between fuel consumption and CO2 effects on the one hand and non-CO2 effects on the other hand needs to be considered. In this context, an implementation of measures targeting both CO2 and non-CO2 effects is typically associated with a cost increase limiting fast implementation from an economic point of view. Therefore, this study investigates implications a market-based policy scheme to support different measures in their implementation. An extension of the EU ETS to account for non-CO2 effects in terms of CO2 equivalents is modeled and resulting climate mitigation potentials and operating cost changes can be analyzed. On this basis, an efficiency comparison of different technical and operational measures can be performed.

ABSTRACT. This study examines the airline network effects of possible non-CO2 climate effect pricing mechanisms under the European Union Emissions Trading System (EU-ETS) on air transport demand and ticket pricing. It is focusing on two regulatory scopes: full coverage of all flights including those in and out of the EEA (European Economic Area) and a reduced scope limited to intra-European routes. Ticket price adjustments are analyzed using pre-calculated trajectory optimizations that account for varying weather conditions, particularly regarding the formation of contrails. Contrail-induced cloudiness significantly enhances non-CO2 climate effects increasing costs to different extents depending on the geographical scope of the regulation. Based on these cost increases, demand changes are assessed through passenger preference models that incorporate intra-market dynamics, capturing the competitive and behavioral responses of passengers across different market segments. This analysis provides valuable insights into the interplay between environmental policy and market behavior. It particularly includes mitigation strategies for non-CO2 impacts, airline pricing strategies, and passenger behavior, offering guidance for the design of effective and equitable climate mitigation measures in aviation by quantifying the demand and network effect of these measures.

ABSTRACT. This paper introduces a comprehensive framework to analyze energy consumption of an electric bus (EB) on fixed transit routes by dividing route geometry into measured segments. Urban bus transit routes of mixed characteristics are examined to understand the benefits of aggregating data spatially for electricity estimation. We incorporate environment specific data such as road gradient, weather and seasonal variables, and time of day with the base data. Real electric bus telemetry is sorted into containers of desired length, route geometry and shape, and we examine its application as an input for traditional and machine learning (ML) models to calculate energy used with aggregated data strategies. Results allow flexibility for operators and researchers to analyze a variety of existing routes or future routes without facing the challenges of time variance between trips of the same route.

ABSTRACT. The interactions between transport and the environment are crucial for understanding the dynamics of pollutant concentration and its contribution to ambient air quality. Roadway and traffic conditions prevailing in urban road networks provide a solid foundation for modelling the distribution of pollution concentration over space and time. These factors also help quantify the extreme influence and exposure of pollutants, such as PM2.5, at critical locations like urban intersections. This study validates the applicability of Extreme Value Theory (EVT) for identifying critical events and locations with high concentrations of particulate matter (PM2.5) using Generalized Extreme Value (GEV) distribution. Rather than relying on selected static monitoring stations of PM2.5, the authors deployed a well-calibrated mobile sensor mounted on a test vehicle, a motorized two-wheeler (chosen for its susceptibility to pollutant exposure). Accompanying this, there were also other vehicle motion and position sensors were used. The equipped vehicle was used for seven days to collect dynamic variations in PM2.5 concentrations along a 6.8 km urban arterial corridor at various periods. The corridor features diverse land uses and associated activities. In addition to traffic variables such as travel times, continuous instantaneous vehicle speeds, acceleration and delay, the study also records metrological variables, including temperature and relative humidity. Significant variability and peaks in PM2.5 were observed concentrations, particularly near urban intersections and locations and commercial and industrial land-use areas. A GEV distribution was found to be the best fit for explaining the variability of PM2.5 concentrations through its parameters (location μ, scale σ, and shape ξ), across 34 blocks of 200 m each. The extreme events associated with exceeding critical PM2.5 concentrations were then predicated using a block-maxima-based EVT approach. This method aims to capture the potential stationarity and non-stationarity of PM2.5 concentrations, as well as relevant independent variables and covariates, using a maximum likelihood point estimate. It is expected that the insights derived from this study can be effectively adopted to evaluate well-designed policies and traffic measures, including interventions in road geometry and traffic control schemes.

ABSTRACT. This study presents a reinforcement learning (RL) based framework involving the integrated use of ramp metering (RM) and variable speed limit (VSL) control towards the ultimate aim of mitigating traffic congestion and emissions. Traditional freeway traffic control strategies often fail to adapt dynamically to evolving traffic conditions, resulting in suboptimal performance. The proposed framework seeks, through simulation, the optimal setting of VSL and RM actions by leveraging RL. The learning-based architecture we have designed is trained and tested using data from a hypothetical freeway network piece and synthetic demand profiles. The performance of the framework is evaluated by considering multiple traffic demand levels and connected and automated vehicle penetration rates.

ABSTRACT. This study proposes an enhanced cellular automaton that provides labeled data based on the three-phase traffic flow theory. In traffic analysis using machine learning techniques, many studies face challenges related to data that lack completeness, particularly due to the difficulty of clearly identifying the traffic phases. To address this issue, the proposed Traffic Cellular Automaton uses microscopic information to provide labeled data based on each traffic phase. Additionally, the study demonstrates the effectiveness of Graph Neural Networks in utilizing this data for traffic classification. This approach ensures clear distinctions between traffic phases, each characterized by unique macroscopic properties, and provides a foundation for further exploration of influential factors in traffic models.

ABSTRACT. Freeway control systems (FCS) improve traffic safety and optimize traffic flow by influencing road users according to the traffic situation. FCS detect traffic conditions using a control algorithm and automatically display variable speed limits, overtaking restrictions, and warnings on variable message signs along the freeway. The integration of FCS into traffic flow simulations is crucial for realistic freeway modeling, enabling, for example, studies of driving behavior, autonomous driving, and further FCS improvement. However, manually implementing FCS in simulation is highly labor-intensive and often results in case-by-case solutions due to local specifications. Additionally, the lack of comprehensive and up-to-date documentation on the implemented control algorithms poses further challenges. This study seeks to address these challenges by leveraging neural networks to automatically reconstruct the control algorithm from traffic and display data, learning the relationships between traffic conditions and displayed messages. First results show a good accuracy of reconstruction and already lead to improved simulations compared to the use of a static speed limit.

ABSTRACT. Traffic management is a critical issue due to its significant environmental and health impact. To address this challenge, we propose a re-routing algorithm based on the integration of Variable Message Signs (VMS) to optimize traffic flow and reduce congestion. The core concept is to develop a system that combines real-time optimization with predictive scenario simulation to effectively manage traffic disruptions, road closing, and other unforeseen events. The proposed methodology includes traffic data analysis to identify network links prone to congestion or inefficiencies. Travellers’ behavior is modelled using the User Equilibrium (UE) framework. Then, the algorithm evaluates potential interventions, considering VMS nodes as critical points for redirecting traffic. Results show that this adaptive traffic management strategy ensures efficiency and effectively supports planned interventions through scenarios evaluation. This work has been partially supported by CN MOST program (PNRR Sustainable Mobility Center, grant n. CN00000023, CUP B43C22000440001).

ABSTRACT. Drone observations on freeways offer exceptional insights into spatio-temporal microscopic traffic behavior, as the drone video captures a 600 m freeway segment over 20 minutes per drone flight in parallel with a data resolution of 0.1 seconds. Especially in dense and congested traffic the data analysis of lane changing on the different freeway lanes offers extended insights into the driver behavior. The approach defines the criticality of the vehicle maneuvers in the different traffic phases as defined in Kerner’s three phase traffic theory based on freeway segments in Germany, China and the United States. The analysis demonstrates the high-risk human drivers take while overtaking and passing activities which are and will be almost impossible for assisted or automated driving systems. This study shows that around 40 % of cars drive with a net gap smaller than 1.8 s and are, therefore, classified as risky. In addition, we analyze lane changing maneuvers and show that only 25 % of lane changes are classified as safe for surrounding vehicles. Due to our suggested hypothetical model for a lane change assistant, we can increase 50 % of all lane-changing maneuvers into a risk-free zone.

ABSTRACT. INTRODUCTION This paper focuses on non-driving-related tasks (NDRTs) at autonomous SAE Level 3, conditionally automated driving (SAE, 2021). This level enables the system to control driving under exactly given conditions. The driver is not requested to hold the steering wheel or monitor traffic. However, drivers must stay prepared to intervene when the system asks the driver to take over (takeover request – TOR = transition between autonomous and manual steering modes). One important related benefit of autonomous vehicles Level 3 (L3) is that the driver will have the space to engage in activities other than driving. This raises the question of what activities are appropriate for this type of driving and how they may affect road safety.

The concept of NDRTs originally came from the field of manually driven vehicles. These activities are not directly related to driving and are mainly linked to satisfying the driver's needs (e.g. air conditioning, radio, music, traffic information, navigation). In automated driving, drivers are expected to engage in non-driving-related tasks. In the conducted studies, the effect of non-driving-related tasks on performance during a takeover is studied. In general, studies point to an increased workload caused by NDRTs and a decrease of situational awareness, which contributes to drivers needing longer time to reorient themselves to the driving task.

A key theory for describing the influence of activity on the process of taking control is Wickens' multiple resources model (MRM) (e.g., Wickens, 2008). This theory describes the encoding and processing of information in the context of so-called resources. According to this model, all tasks use certain resources and if these tasks are performed at the same period, there is a possibility of interference occurring when two tasks use the same resources. On the other hand, no interference occurs in situations where the tasks are not similar.

NDRTs are classified from different perspectives. One of the most significant features is the sensory modality of the activity. The type of sensory modality has a significant effect on the ability to take control back. Auditory activities have a lower negative influence on driving performance during takeover than visual activities. Another important characteristic that affects the safety of taking control is holding an object in the hand. Research shows that having an object in the hand and putting it down leads to a longer time to take over the wheel. The important criterion for classifying activities is the degree of cognitive load (activity demands). All types of workload have been shown to negatively impact the ability to take control by affecting driving performance.

Within the literature are relatively separate criteria describing the riskiness of a particular type of NDRT. In the research conducted, one of the research objectives was to establish a systematic procedure that would allow for assessing different types of tasks. The novelty of our approach can be summarised as a unique rating system using several criteria which leads to the determination of suitability at Level 3. The background to this approach and the assessment procedure are also described below, to provide a model for assessing NDRTs for Level 3.

METHODS The Risk Assessment Model of non-driving-related tasks was developed based on several cornerstones. The first was a thorough theoretical search, which included results from studies conducted on the impact of activities and theoretical assumptions for performing the activities (Wickens' resource theory). The second one was a simulator study using a full motion truck simulator, performed by authors (Horáková & Krejčí, 2024) in 2023. The main focus of this study was to evaluate the effect of different types of NDRTs on takeover performance. The sample consisted of 31 participants (professional truck or bus drivers). The results show that takeover is aggravated by handling objects with, a higher visual and mental load. Takeover reaction times differ between activities, some statistical significance has been confirmed. These results verified some assumptions based on the analysis of studies and theory.

RESULTS The model describes the process of assessing NDRTs. Firstly, three key areas for assessment have been identified as playing an important role in the takeover and need to be evaluated. Then, key characteristics entering the assessment model were identified. Based on the default assumptions, three topics were identified as key to assessing the impact of the activities on the process of taking control: the type of input modality information that is perceived, the mental processing of information that is related to the activity being performed and the physical circumstances of the activity execution. By analogy to these three headings, three key factors have been developed that enter the evaluation process: the first factor of sensory-visual limitation, the second factor of total high demands when performing activities, and the third factor of motoric limitation.

Based on the three main headings, the individual characteristics (features) were then defined, which can be used to evaluate the factors. These features include for example sensory modality as the key in the perceptual process of performing an activity (visual or auditory modality), mental demands assessed as the amount and complexity of cognitive stimuli (low-medium-high scale), holding an object in hand as a physical restriction (yes or no) and other similar characteristics.

The process of evaluation of a non-driving-related task has three steps. The first step is qualitatively assessing the key characteristics of the non-driving-related task. These qualitative data are then assigned quantitative values, which are used for the next evaluation step. The second step is determining the numerical level of all three factors that enter the overall evaluation of the non-driving-related task. The resulting levels are determined by assessing all characteristics, with the numerical values at the intersection of the relevant row and column. The third and final step is to carry out an overall assessment of the non-driving-related task.

The overall assessment is an integration of the specified levels of all three factors. This rating then indicates whether the activity is suitable for this level of automation. Specifically, this is classified into 3 categories in terms of suitability for L 3 based on the evaluation: safe NDRT, risky NDRT, and ambiguous NDRT. (Note: the specific model, using the individual qualitative and quantitative parameters as well as the overall graphical representation of the model and examples for the assessment process will be described in more detail in the full paper).

CONCLUSION The full paper will provide a thorough discussion section comparing our approach with other developed systems. Its strengths and limitations will be highlighted. It will also be compared with how the performance of activities is currently set up in practice (mostly on the side of autonomous vehicle manufacturers). The benefits of the Risk Assessment Model are threefold. Firstly, for a safe approach in the preparation law environment for L3; secondly, for high-quality awareness campaigns; and lastly, for manufacturers of L3 vehicles to fine-tune the setup of HMI.

REFERENCES: Horáková, M. &, Krejčí, L. (2024). Effects of Non-Driving Related Tasks on Driving Performance after Take-Over Request in Automated Vehicles (Level 3). European Transport Conference, 18-20. September 2024, Belgium – Antwerp. https://aetransport.org/past-etc-papers/conference-papers-2024?abstractId=8319&state=b SAE International (2021). Taxonomy and definitions for terms related to on-road motor vehicle automated driving systems. https://www.sae.org/standards/content/j3016_202104 Wickens, C. D. (2008). Multiple resources and mental workload. Hum. Factors 50(3), 449–455. https://doi.org/10.1518/001872008X288394

ABSTRACT. Advanced Driver Assistance Systems (ADAS) and Automated Driving Systems (ADS) promise to enhance driving safety and comfort, yet their complexity often leads to mode confusion, where drivers misunderstand or remain unaware of a vehicle’s operational mode. We conducted a targeted literature search and analyzed a survey dataset (N=838) to identify the contextual, vehicle-related, and driver-specific factors most likely to shape drivers’ mental models of ADAS. Preliminary results suggest that age, driving distance, and car ownership are particularly salient in predicting “I don’t know if I use it” responses to features such as Adaptive Cruise Control, Lane Keeping Assistance, and Lane Departure Warning. Building on these insights, we propose a conceptual model that highlights key factors underpinning driver mental model misalignment with ADAS functionality. Future work will systematically validate this model through a dedicated questionnaire, aiming to minimize mode confusion and foster safer ADAS usage.

ABSTRACT. The use of drones for freight delivery has been extensively studied in the context of last-mile logistics, with existing literature underlining potential benefits across several dimensions, from economic efficiency to reductions in CO₂ emissions. But can these advantages also extend to longer routes? By relying on real-world data on freight transported between distribution and sorting centers of one of Italy’s major freight operators, we propose different modeling frameworks to assess the feasibility of using drones for middle-mile logistics. Specifically, our proposed models address both drone location and flight scheduling for a single freight operator, with the aim of maximizing drone utilization as a replacement for current ground-based middle-mile operations within a given fleet size. We introduce three different model specifications, namely Maximization of Freight Transported (M1), Maximization of Ground Movements Replaced by Drones (M2), and Minimization of CO₂ Emissions (M3). To practically implement the proposed optimization models using real-world data, we considered current freight demand of a typical working day and schedule, along with specific operational assumptions. These include the technical limitations of drones, based on drones currently used for similar purposes, as well as constraints related to battery charging and regulation. The results reveal that the current technical characteristics of drones are insufficient to fully replace ground-based freight transport. A key limitation is that the maximum flight range of a drone depends on its weight, leading to significant inefficiencies given the structure of the middle-mile network. By applying our models to the prototypical day of operations we found that the results under the different objectives are quite aligned. Nevertheless, we also observed peculiarities when considering the various objectives. In the two analyzed networks, the maximum achievable CO₂ emission savings range from 5.1% with 10 drones to 11.6% with 150 drones. The proportion of ground movements replaced reaches a maximum of 28% with 10 drones and 51.3% with 150 drones, while the largest proportion of diverted freight demand is 13.9% and 34.6% for 10 and 150 drones, respectively. Overall, we conclude that further technological advancements are required before drone operations will be possible for middle-mile operations.

ABSTRACT. Abstract Multimodal mobility hubs are emerging as an effective solution for integrating various transport modes and reducing reliance on private cars. Their successful design and implementation require a thorough understanding of factors affecting user preferences and behaviors concerning the hubs and the transport modes they offer. This study used a stated choice experiment to examine travel preferences among car drivers in the Netherlands. A differentiation was made between urban and regional hubs in measuring travellers preferences. A comprehensive set of attributes incorporating hub characteristics, travel mode attributes, and contextual variables were included in the experiment. The results of an estimated error component mixed logit models show that private cars remain the preferred mode in both types of hubs. Factors like sunny weather, shorter travel distances, and free parking at hubs increased hub attractiveness. Buses and trains were preferred in urban and regional hubs, respectively, compared to shared e-bike, share e-scooter and shared car. Preferences differed by age, education, driving frequency, and familiarity with shared modes, showing distinct patterns across different type of hub. 1. Introduction The concept of multimodality is gaining increasing attention as a means to shift toward more sustainable and efficient transportation systems. It refers to the integration of multiple transport modes, such as public transit, cycling, shared vehicles, and on-demand services, into a cohesive and flexible system, enabling users to seamlessly switch between options depending on their needs. This approach is supported by both physical solutions, such as park and ride facilities and multimodal transport hubs, and digital solutions, such as Mobility as a Service apps for route planning, real-time updates, and payment systems. Multimodal hubs, in particular, play a central role by providing easy access to multiple transport options in one physical location, bridging the gap between private car use and public or shared transport, and ensuring smoother transitions between modes. A key barrier to the widespread adoption of multimodal hubs lies in addressing the needs and preferences of frequent car drivers, who are often reluctant to give up their private vehicles. For many car users, the car offers a unique combination of convenience, comfort, freedom and flexibility, offering a door-to-door solution that other transport modes may not easily replicate (Kent, 2014; Hunecke et al., 2021). While multimodal hubs present promising alternatives, the challenge is making these options equally appealing. If alternatives do not match or exceed the comfort, flexibility, convenience or cost of private car use, drivers are unlikely to consider them. To attract car users to multimodal hubs, it is crucial to enhance both service levels (e.g. availability of transport options, their frequency, accessibility) and the range of facilities offered with reasonable prices. By incorporating diverse services, hubs can become lively spaces where people can conduct various activities beyond just traveling. As such, the disutility of traveling can be offset by the gained utility from engaging in activities. A component which a private car cannot offer. For example, cafes or co-working spaces can offer comfortable waiting areas, while supermarkets or parcel pickup points enable passengers to run errands. This integrated approach can reduce friction and make multimodal trips more practical and attractive, offering convenience and flexibility to users accustomed to the door-to-door convenience of private car ownership, and can encourage people to view hubs as part of an integrated journey (Roukoni et al, 2023). This research fills important gaps in the existing literature by providing a comprehensive evaluation of multimodal hubs’ potential to facilitate mode switching. Unlike previous studies focusing on hub location (Blad et al., 2022), user segmentation in neighbourhood hubs (van der Meer et al., 2023), or electric shared mobility modes in eHUBS (Liao et al., 2024), this study offers novel insights into the role of multimodal hubs in both urban and regional contexts. It investigates how public transport and shared mobility modes, integrated within these hubs, can replace private car use and explores the factors driving mode switching, with a particular focus on innovative hub features like social extended reality offices. Using a stated choice experiment distributed among frequent car drivers, the study provides unique insights into mode preferences and the influence of hub amenities. Two types of hubs are explored: (i) urban mobility hubs in city centers, offering alternatives for shorter travel distances, and (ii) regional hubs in less populated areas, offering alternatives for longer journeys. By examining these hubs separately, the research sheds light on how their characteristics and services influence user behaviour and whether differences exist between the two types of hubs when it comes to users preferences and behavior. 2. Materials and methods The study used a stated choice experiment to investigate the adoption of multimodal mobility hubs among frequent car drivers in the Netherlands. Data were collected via an online survey distributed through a market research company. The survey was structured into several sections, including screening for car ownership and driving frequency, questions about private and shared transport mode usage, a stated choice experiment, attitudes toward multimodality and the environment, and socio-demographic information. Two experimental setups were considered: urban hubs and regional hubs. Both setups shared the same number of alternatives, attributes, and levels. An orthogonal fractional factorial design of 64 rows, blocked into 8 subsets, was used. Respondents were randomly assigned to one block and one hub type. In the choice task, respondents were presented with a hypothetical travel scenario in which they imagined making a trip of a certain distance for a specific purpose under given weather conditions. They were then asked whether they would travel by private car or via the multimodal hub. If they opted for the hub, they selected a transport mode from the available options: shared e-bikes, e-scooters, and shared cars were available at both hub types, while buses were included in urban hubs and trains in regional hubs to reflect their typical use. Attributes were divided into contextual factors (e.g., weather, travel purpose, secondary activities), private car attributes (e.g., car travel distance, parking costs), mobility hub attributes (e.g., distance to hub, transfer time, facilities), and transport mode attributes, which refer to the modes offered within the multimodal hub (e.g., tariffs, travel times). Travel times were calculated using average speeds for different distance ranges, and travel costs were calculated based on distances and market rates. After cleaning the dataset, 441 valid responses were obtained for the urban hub, contributing 3,528 choice observations, and 447 responses for the regional hub, which led to 3,576 observations. No significant demographic differences were found between the two samples. Due to the screening criteria, all were frequent car drivers who reported to drive at least once per week and own at least one car. This implied that the sample was not fully representative of the Dutch population. It showed nearly equal gender representation, with more participants being older and having higher incomes, while there were fewer with lower incomes and lower level of education. Most individuals owned a single private gasoline vehicle and were not familiar with shared transportation options. The data were analysed using error components mixed logit models, developed separately for urban and regional hubs, to examine respondents’ choices between private car use and mobility hub, as well as their selected mode of transport if choosing the hub. These models accounted for panel effects to capture repeated choices made by the same respondent across different scenarios. An error component was included for shared e-bike and shared e-scooter to account for potential unobserved correlations between these modes, as they are both lightweight, shared, and individual transport options that may appeal to similar user preferences. Socio-demographic and travel-related characteristics were included in the models to better understand how individual attributes, such as age, income, and car usage patterns, influence preferences and decision-making. Only significant variables were retained in the final models to ensure robustness and interpretability of the results. 3. Results The results reveal several key patterns in individual preferences. In both urban and regional hubs, the alternative-specific constants demonstrate a strong inclination towards private cars which is also confirmed by descriptive data analysis. Within hubs, bus and train were preferred for urban and regional settings, respectively. The significant standard deviations of the alternative specific constants indicate the existence of significant variation across individuals, highlighting the diversity in preferences toward the available modes. Moreover, the significant error component coefficients confirm the correlations among the micromobility alternatives. In both urban and regional hubs, cost sensitivity significantly influences decision-making. High parking fees at destination strongly discourage car use, while free parking at the hub and lower tariffs for public transport and shared modes encourage their use, although to different extent. This confirms that pricing, whether for parking or tariffs, is a crucial determinant for mode selection. The models' estimates for both types of hubs further indicate that travel distance substantially influences travel choice in both contexts. Longer origin-to-destination distances tend to favour private car use, implying that private cars are perceived as more convenient for covering extensive distances. However, when shorter distances to mobility hubs are available, individuals are more likely to choose modes that involve traveling via these hubs. This indicates that proximity to a hub increases its attractiveness as a transit point. The distance from the hub to the final destination also influences the mode choice, with varying impacts depending on its location relative to the overall trip and type of mode. Weather conditions also prove to be a significant determinant, especially for bike and scooter usage. These modes are chosen more frequently in favourable weather conditions, which highlights the importance of considering weather as a contextual factor in predicting mode choice. In addition, the primary trip purpose plays a critical role. When commuting for work, individuals tend to choose private car. An important difference between the urban and regional hub models lies in the influence of available functions at the hubs. Lastly, preferences varied based on factors such as age, education, driving frequency, and familiarity with shared modes, highlighting distinct patterns between urban and regional hubs. For example, younger respondents and those who use the car less frequently tended to be more open to multimodal travel, particularly in urban settings. 4. Conclusion The research provides key insights for urban planning and transportation policy, focusing on integration of public transportation and shared mobility. The findings highlight that strategic placement of multimodal hubs is essential and the presence of diverse facilities can enhance hub appeal and adoption. Economic incentives, like reduced parking fees at hubs, can encourage alternative transport use. The study's limitations include the hypothetical nature of the experiment and unfamiliarity with multimodal hubs.

ABSTRACT. The motorhome tourism model, which includes motorhomes, caravans or camper-type vehicles, has also established itself as a preferred travel option for those seeking flexibility, contact with nature and a more personal travel experience less constrained by traditional tourist infrastructure. This model is characterised, firstly, by the ability it offers travellers to move around autonomously, allowing for modifications to routes, itineraries and length of stay according to personal needs and preferences. Secondly, recreational vehicles allow users to avoid large tourist concentrations, a feature that has become particularly attractive in the wake of the 2020 lockdowns due to the COVID-19 pandemic. During this period, interest in motorhome tourism increased significantly worldwide, partly due to mobility restrictions and the need for social distancing (Mundet, Grijalvo and Marin, 2024). The growth of motorhome tourism has also had a significant economic impact on associated products and services, such as rental services, the sale of motorhomes and specialised equipment (Mason, 2019). Several studies indicate that the expansion of the motorhome market has boosted complementary industries, such as vehicle maintenance, specialised service stations and camping businesses. In Spain, the growth of the sector has also boosted demand for infrastructure dedicated to overnight accommodation for these vehicles, in the form of specially designated and signposted areas for motorhomes. This trend reflects a shift in the tourism industry, with travellers increasingly seeking independent experiences, often geared towards ecotourism and the preservation of natural environments (Alkan, 2021). The economic and social impact of this increase in motorhome tourism is evident in the increased demand for related services, such as maintenance and specialised camping facilities. The sector has not only influenced tourism, but has also contributed to a broader market that includes vehicle maintenance and service stations (Eurostat, 2022). In addition, the shift towards mobile tourism has fostered a broader interest in more flexible travel options and destinations that offer a combination of green and sustainable experiences (Espiner et al., 2023). Furthermore, the growth of motorhome tourism has given rise to the need to better understand the different categories of motorhome travellers. Research by Marques and Rodrigues (2024) suggests that there is a typology of motorhome travellers in Europe, including ‘enjoyers’, ‘seekers’ and ‘vacationers’, each segment with different preferences and behaviours when choosing their travel destinations. Also, the new data analytics technics has emerged in the last years and its application has increased in many fields. In this context numerous studies have analysed prices in the accommodation sectors, using various econometric and machine learning models to predict rates based on location, services and additional amenities. Machine learning techniques such as XGB, Random Forest and OL models are commonly applied in pricing studies in different industries, including tourism, transportation and housing (McHugh et al., 2020; Zong, 2023; Huang, 2024). For example, machine learning models have been used to predict rates in different tourism sectors by evaluating variables such as location, services offered and demand (Fang, 2023). A study by Bukvić (2022) on pricing strategies in the vehicles markets highlights the importance of these factors in influencing consumers' willingness to pay. In addition, models such as ordered logit are often used to classify prices into ranges, as they provide an understanding of the distribution of prices according to different levels of service or demand (Williams, 2016). Despite the significant growth of motorhome areas in Spain, with more than 800 operational sites in both the public and private sectors, determining optimal pricing factors remains a challenge. Only, specific research has been carried out to understand the distribution of motorhome areas by type of management (public vs. private) and their spatial positioning, often revealing correlations between location, services offered and pricing structures. For example, the study by García Madruga (2024) on price optimisation in tourism highlights the importance of service offerings such as electricity and water connections in determining the final price for customers. The variety of characteristics and services offered by these areas adds complexity to the pricing process, making it necessary to develop a methodology for determining rates and filling a gap in the literature. Although existing studies have explored pricing strategies in the accommodation sector, an approach that combines data analysis and pricing models specific to the motorhome tourism sector has not been fully addressed. This study addresses this gap in the literature, and proposes the integration of machine learning (ML) techniques and an ordered logit (OL) model to identify and optimise the factors that influence rate setting.

ABSTRACT. The successful implementation of on-demand feeder services requires identifying urban areas where these services can maximise user convenience and integrate seamlessly with existing public transport, enhancing efficiency and sustainability. However, policymakers lack tools to determine the most promising locations for such services while considering all major urban transport modes, including private cars. To address this, we refine our utility-based simulation framework, which evaluates urban areas for on-demand feeder bus services by assessing key performance indicators under uncertain demand. The framework considers two travel options: an integrated trip (feeder bus to a hub, then public transport) and direct public transport. We extend this by incorporating private car usage into the mode choice set for a more comprehensive mobility analysis. Using a nested logit model, we capture hierarchical decision-making and assess the potential of feeder services to attract car users to the 'Public Transport Nest' (public transport and integrated feeder modes). With this approach, we provide a detailed assessment of the feeder service attractiveness and its potential to reduce the car dependency in urban areas using Krakow (Poland) as an example.

ABSTRACT. Sustainable mobility has become one of the main targets both in urban planning and environmental policies. As cities and metropolitan regions grow, new mobility patterns present significant challenges but also offer substantial opportunities to reduce the negative externalities generated by transport. While most urban travel involves medium- to short-distance trips, motorized transportation continues to dominate as the preferred transport solution, even for short distances that are compatible with other solutions. Motorized transportation significantly contributes to critical urban issues such as air pollution, greenhouse gas emissions, traffic congestion, and noise pollution (Gavanas, 2024). These problems threaten both ecological balance and public health, affecting many aspects of urban living. Additionally, the "barrier effect" caused by urban road infrastructure fragments communities, discourages the use of non-motorized transportation, and further promotes car dependency (Maciorowski & Souza, 2018). For intra-urban travel characterized by short distances, the bicycle is the most sustainable (and in most cases, also efficient) means of transport. Bicycles do not generate air and noise pollution, cause less traffic congestion, and promote physical health (Maltese et al., 2021). Studies indicate that cities with higher rates of cycling tend to enjoy better air quality, lower healthcare costs, and improved overall well-being. Beyond these benefits, cycling also supports urban economies by reducing road maintenance costs, decreasing fuel consumption, and boosting local businesses through enhanced access to pedestrian- and bike-friendly areas. Theoretically, mediterranean countries, due to the favorable climate, are a privileged location for the use of bicycle. However, several infrastructural and cultural aspects limit its adoption. In Italy, for instance, despite the presence of 22 million bicycles in use, cycling culture remains limited by ingrained cultural barriers, inadequate infrastructure, and insufficient awareness of the social and environmental benefits of cycling (Brčić et al., 2016; Privitera, 2020). Structurally, many urban residents perceive cycling as unsafe or impractical due to poorly designed bike paths, lack of connectivity, or concerns about theft. Policy barriers, such as fragmented cycling networks and governance challenges, further impede the shift toward sustainable mobility (Bardal et al., 2020). As a result, the bicycle is not one of the main options in modal share of most Italian and Mediterranean cities. These obstacles cannot be overcome without consistent investment and integrated strategies. Infrastructural development is a requisite for improving the use of bicycles, as evidenced by studies highlighting the role of dedicated cycle lanes in promoting cycling culture (Pucher et al., 2010; Aldred et al., 2017). The experience of cities like Copenhagen and Amsterdam, which invested heavily in efficient cycling networks, demonstrates the success of such infrastructure in increasing bicycle usage (Fishman, 2016). However, infrastructural development alone is not sufficient to guarantee the widespread adoption of cycling as a transportation alternative. Although bicycles are increasingly used for short- and medium-distance travel, resistance remains due to personal preferences or a lack of awareness regarding the benefits of cycling (Heinen et al., 2011). Many individuals remain unaware of the substantial positive impact that cycling can have on health, such as reduced risks of cardiovascular diseases (Woodcock et al., 2009), environmental sustainability, and urban livability (Götschi et al., 2016). To address this, supplementary measures aimed at educating and motivating potential users are essential. One promising approach is the development of tools like calculators that estimate the advantages of switching to bicycles. Such calculators could quantify environmental, economic, and social benefits, such as reductions in CO2 emissions, personal fuel expenses, and improvements in fitness metrics, addressing critical knowledge gaps (Macmillan et al., 2014). These tools could empower individuals to visualize the tangible benefits of cycling, potentially leading to lasting behavioral changes. Despite the potential utility of such tools, they remain underdeveloped and are not widely available to (potential) cyclists, underscoring the need for innovative solutions in this domain. To cover this gap, this contribution presents the outcomes of the project “CICLO!”, which develops a calculator that may be useful for potential bike users.

ABSTRACT. The rapid urbanization and growing awareness of the need for ecologically sustainable mobility have led to an increasing shift towards active transport modes. The complexity of urban mobility planning requires innovative solutions that go beyond traditional time and distance metrics. This research presents the design and development of a multicriteria route planner app for mobile devices, focusing on active transport modes such as walking and cycling. Unlike existing solutions, this application integrates additional variables related to safety, accessibility, comfort, and environmental quality, providing a more holistic approach to urban mobility planning. A key objective of this study was to evaluate the feasibility of incorporating crowdsourcing as a mechanism for updating real-time route variables. The research explored the viability of engaging users in providing information on road conditions, accessibility issues, and safety concerns. Additionally, a user-centered methodology was applied to assess the usability and efficiency of the developed prototype, ensuring that it met the expectations and needs of potential users. The study followed an exploratory methodology comprising four phases: literature review, user needs assessment through surveys, prototype development, and usability evaluation through user testing and feedback. The literature review identified a gap in existing route planners, which predominantly focus on shortest path algorithms, often disregarding key factors affecting route choices for users of active transport modes. Several user surveys and interviews revealed that urban commuters prioritize attributes such as surface quality, lighting, crime levels, and air quality when choosing walking and cycling routes. A prototype incorporating these criteria was designed following human-computer interaction principles and heuristics to optimize usability. The app integrates a ranking system to incentivize user participation in data updating, reinforcing the crowdsourcing approach as a viable solution for maintaining accurate and relevant information. To validate the developed solution, a usability evaluation was conducted through real-world testing scenarios where participants simulated navigation using the app. Results indicate that users found the interface intuitive and the additional route information highly valuable for planning safe and comfortable journeys. Moreover, the integration of user-generated updates proved effective in maintaining route information accuracy. The weighting system within the app, which assigns relative importance to different route criteria, was refined based on user preferences and feedback, ensuring optimal alignment with commuter expectations.

The findings highlight the potential of a multicriteria route planner for improving urban mobility by offering more personalized and informative travel options. By leveraging crowdsourcing, this application ensures continuous data updates while fostering community engagement. Furthermore, the study highlights the importance of integrating real-time data, such as traffic conditions and environmental quality, to enhance route selection and sustainability. This research contributes to the growing field of intelligent mobility solutions, advocating for data-driven urban transport planning that accommodates diverse mobility needs. Future work will focus on expanding the crowdsourcing model, integrating real-time sensor data, and developing multimodal route options that combine active transport modes with public transit systems. Collaboration with municipal authorities and transport agencies will be essential to scale and deploy the proposed solution in real-world urban settings, ultimately promoting sustainable and efficient urban mobility.

ABSTRACT. The concept of social sustainability is multifaceted and continues to be a topic of considerable debate, particularly with regard to its indicators and applications. The existing literature on social sustainability in transportation is limited with regard to its definition and assessment. Moreover, current approaches often neglect the subjective perceptions of individuals, prioritizing objective aspects, such as quantitative indices, in the exploration and evaluation of social sustainability in transportation. This approach often overlooks the psychological characteristics that are integral comprehending and assessing of the social sustainability in transportation systems. To address this gap, this research proposes the development of two questionnaires designed to assess social sustainability based on citizens' experiences. The objective of these tools is to quantify social sustainability in urban mobility contexts, taking into account both objective indicators and subjective dimensions. By capturing individual perspectives, the questionnaires will facilitate the design of mobility solutions that are inclusive, tailored to user needs, and capable of promoting sustainable mobility practices.

ABSTRACT. The integration of modern transport infrastructure into heritage cities presents significant challenges, particularly in balancing the preservation of historical sites with the introduction of new mobility solutions. This paper explores the implementation of cycleways in historic urban areas, investigating how cycling infrastructure can coexist with the cultural and architectural integrity of these cities. Through a mixed-methods approach combining semi-structured interviews with urban planners, policymakers, and heritage experts, case studies of successful and unsuccessful cycleway implementations, and a policy analysis framework, this study identifies the key challenges and strategies for integrating cycling infrastructure in heritage contexts. Key challenges include community resistance, regulatory constraints, and the potential for damage to historical landmarks. Successful strategies include collaborative planning, innovative design solutions, and policy innovation. The study applies a systems thinking framework to identify leverage points that can lead to sustainable and culturally sensitive outcomes, such as enhancing urban policies and fostering stakeholder engagement. The findings offer actionable recommendations for urban planners, policymakers, and transport engineers to develop cycling infrastructure that supports both mobility and heritage preservation. This research contributes to the development of more informed, balanced transport planning strategies that ensure the sustainable integration of cycleways in heritage cities, promoting both sustainable mobility and the conservation of cultural heritage.

ABSTRACT. To address the negative effects of private car usage, there is a growing urgency to promote sustainable transport alternatives, aimed at reducing the number of private cars on the road and encouraging cleaner and more efficient transport modes to mitigate these negative impacts. An interesting focus has been put on more integrated systems of transport envisioning seamless journeys using different modes of transport through the concept of “intermodality”, which consists on combining the use of different transport modes within a single trip. The premise is that the integration of different transport modes could provide a more sustainable and competitive transport option than using those modes individually. One example is the modal combination between public transport (PT) for longer distances and the bike for the first- and last-mile of trips, with PT significantly increasing the speed and coverage area of cycling and the bike offering flexibility and accessibility. Intermodal mobility can significantly reduce CO2 emissions compared to trips reliant solely on private vehicles, improving accessibility and reducing travel times as well as making public transport a more attractive option for commuters. However, despite the potential benefits of intermodal mobility, research in this area is still in its early stages. Only a few studies have so far explored intermodality in everyday mobility, with research focusing on the travel behaviour of intermodal users being especially rare. To gain a comprehensive understanding of intermodality and its potential contribution for sustainable mobility it is essential to delve into the underlying rationale and patterns of intermodal travel behaviour from the user's perspective. Accordingly, this study investigates intermodal travel behaviour using as case study of the city of Porto, a traditional car-centric urban area undergoing significant transformations in its transport system. Despite improvements such as the expansion of its metro system and the introduction of new mobility services like ride-hailing and e-scooter sharing, private car dependency remains high. Through the implementation of a representative mobility survey, this research identifies and characterizes different profiles of intermodal users focusing on four main groups of factors: 1) travel patterns, including trip purposes and modal combinations; 2) socioeconomic and demographic characteristics; 3) the built and natural environment of their main origins and destinations; and 4) their main motivations and barriers for intermodal travel. Given its sustainability potential, this research focuses on users who combine different modes of public transport (e.g., train and bus) or public transport with other modes (including private vehicles, shared mobility and active modes such as walking and cycling) into a single trip. By identifying the most significant factors that influence intermodal users, this study provides valuable insights for policymakers and transport operators, guiding the design of more integrated and efficient intermodal transport systems that support a shift to a more sustainable urban mobility.

ABSTRACT. Understanding the relationship between weather events and highway traffic flow is important for enhancing transportation infrastructure resilience. This study investigates the effects of localized rain events on highway capacity, with a specific focus on traffic breakdowns. Analysis of connected vehicle data under varying rain conditions reveals that traffic breakdowns are most likely to occur at the boundaries of rain zones. Short, localized rain events disrupt traffic more significantly than widespread, persistent rainfall, due to a habituation effect where drivers adjust to prolonged adverse conditions. Unlike previous studies, which broadly examine the impact of rain on traffic flow, this research focuses on the distinct effects of localized rain events, providing a more nuanced understanding of precipitation-induced disruptions. The findings from this study could offer practical guidance for improving traffic management strategies, optimizing road design, and enhancing driver safety during adverse weather events.

ABSTRACT. Cars remain the most widely used mode of transport today. However, in many urban areas, high car usage leads to negative externalities such as congestion, pollution, and inefficient land use.Optimising parking policies in cities is a promising approach to reduce these externalities, though it often involves trade-offs; for example, reducing parking space can increase the time drivers spend searching for a spot.

We present a model to optimise parking capacities in urban areas using a multi-objective framework that simultaneously minimises (1) travel time, (2) distance travelled by car and (3) the number of parking spaces. We address this problem using a bi-level programming framework as parking capacity decisions (upper level) influence driver route and parking choices (lower level), which in turn affect the objective values.Our main methodological contribution lies in enhancing the upper level optimisation through a novel mutation operator, which helps achieve lower objective values.

We apply our model to the city of Delft, the Netherlands, demonstrating that a diverse set of solutions with low objective values can be obtained. Moreover, we show through an example within this case study that our model can help policy-makers assess trade-offs in the conflicting objectives.

ABSTRACT. Public transportation crowding in the morning peak has been a major issue in operation for decades. There have been several attempts to tackle it with off-peak discounts. In this paper, we analyze the potential improvements and propose a pricing scheme based on hourly demand. It offers a distinctive fare difference that encourages passengers to change their departure time and keeps the operator's income stable. An analytical approach with a simple scenario and a case study and simulation with JRE data in Tokyo proves that there is a range of pricing scheme that fulfills the purpose.

ABSTRACT. Bike-sharing (BS) can serve as a first- and last-mile solution and is known as a complementary mode of transportation for the Public Transportation (PT) system. In terms of network design concepts, there are various aspects that can have an impact on the quality of such an integrated system, such as bike availability, PT fleet sizing, and fare for the system. However, in network design problems, these modes are mostly considered in isolation, resulting in a sub-optimal design of the variables. This study aims to optimize the mentioned design variables simultaneously for an integrated system of BS and PT. The objective function of the optimization model is maximizing social welfare. The social welfare function involves operators' surplus, consumers' surplus, and externalities such as climate change costs and health benefits. The model is tested in a numerical example from the literature known as Mandl's network. The results of the algorithm in this numerical example are expected to show a improvement in social welfare compared to conventional, isolation decision-making methods.

ABSTRACT. This study investigates the feasibility and potential impacts of deploying autonomous vehicles (AVs) as a public transport solution in a peri-urban area, using a case study of an existing bus route in Aveiro, Portugal. Five operational scenarios were compared: conventional diesel buses, full electrification, on-demand light AVs, autonomous minibuses, and a hybrid system. The methodology combined real-world data collection, scenario design, route optimization, emission assessment, and cost analysis. Results indicate that light AVs offer significant emission reductions (70-90% compared to electric buses) and cost savings compared to conventional systems. A hybrid model, combining electric buses for peak hours and light AVs for off-peak, emerged as the most promising solution, balancing cost efficiency and environmental impact. The study also suggests potential labor savings by replacing multiple drivers with a single remote operator overseeing several AVs, potentially mitigating driver shortages. However, the study acknowledges the need for realism regarding current AV industry profitability, emphasizing the substantial upfront costs and long-term perspective required for widespread, profitable AV deployment in public transport.

ABSTRACT. The very nature of transportation is being transformed by Autonomous Vehicles (AVs), that was a dream by some point in time. This technological achievement is transforming transportation in unprecedented ways, offering promises of safer roads, increased accessibility and reduced traffic congestion. However, it raises also concerns about cybersecurity attacks (thus concerns about safety), compatibility of AVs with conventional vehicles, suitability of road infrastructure to the AVs and other. Therefore, the main challenge that manufacturers of AVs will need to face is the adoption of such innovation from the consumers. Several studies have been conducted for distinguishing between factors that contribute to autonomous vehicles adoption and factors that prompting resistance for AVs’ adoption. For example, Crisafulli et al., (2025), emphasized factors like personal values, safety, convenience, and trialability as key factors of AV adoption. One of the most known methods for measuring people’s preferences for using AVs, and thus the adoption of AVs, is the stated preference surveys, where respondents are asked hypothetical questions about AV adoption under different scenarios (e.g., cost, travel time, safety measures, etc.). However, studies that have used stated preference may lead to overestimated acceptance rates or misleading conclusions about AV adoption (Carlsson et al., 2018). Therefore, we need to highlight the need for complementary approaches that rely on revealed preference (RP) data, such as travel decision surveys and trip records, which reflect real-world behavior. As denoted from Parekh et al., (2022) there are several challenges that AV technology need to address such as safety concerns, ethical dilemmas and other, which enhance the need to overcoming existing barriers and for achieving a widespread adoption of AVs. Additionally, Shetty et al., (2021) identified concerns about road safety especially that are related with the lack of communication between AVs and conventional vehicles that decreases the possibilities of adoption by frequent commuters. Therefore, overcoming any barriers identified concerning different factors, such as road safety, is essential for the adoption of AVs. However, even overcoming any barriers there is a proportion of the frequent commuters that will persist of not adopting AVs to their mode choices. Thus, it is essential to identify the groups of peoples (personas) that are more willing to adopt AVs. For instance, Janatabadi and Ermagun (2022) revealed that younger, tech-savvy individuals are more likely to accept AVs, while older adults and those with lower technological familiarity express greater skepticism. These findings suggest that public acceptance is not uniform and is shaped by trust, safety perceptions, and prior experience, necessitating a data-driven approach to identify commuter personas and predict how different groups will integrate AVs into their daily routines. By leveraging trip diary data, such as travel decision surveys from San Francisco (USA), this study aims to address the limitations of stated preference studies and provide more accurate insights into AV adoption. Analyzing trip records can reveal how different demographic groups (e.g., older adults, low-income individuals) use existing mobility options, such as micromobility or public transit, and how these patterns might be translated to AV adoption. This approach aligns with the findings of Ghansiyal et al., (2021), which emphasizes the importance of data-driven insights for understanding user behavior and designing AV systems that meet diverse needs. The study will identify distinct commuter personas, based on their demographic and mobility profile, offering actionable insights for policymakers and urban planners to ensure a smoother and more equitable integration of AVs into urban mobility systems. As it was concerned, multimodal-commuters (commuters that use alternative modes for their different purposes of travel during a day), are more willing to adopt AVs In detail, a Binary Logistic Regression model will be developed to analyze the factors influencing multimodal commuters, as they are considered more likely to adopt AVs compared to those who consistently use the same mode for their daily commutes, regardless of trip purpose. By identifying key determinants of AV adoption among multimodal commuters, this model will offer valuable insights into behavioral patterns that can inform urban mobility strategies. Policymakers can utilize the findings to target different commuter segments more effectively, ensuring the promotion of AVs aligns with the specific needs and preferences of diverse commuter personas in San Francisco, USA, based on their socio-economic context. Figure 1, presents the socio-economic context of the participants from the different areas of San Francisco for three indicative years (2015, 2017 and 2019). As can be seen from the figure the travel decision survey covers a sufficient rate of all income and age categories over the entire area of San Francisco, through the three years, and thus was considered as adequate for drawing any conclusions from the analysis, about the adoption of AVs. Furthermore, the travel decision survey includes records of users who have used e-hailing services (e.g., Uber, Lyft) for trips with varying purposes. By analyzing these patterns, the study will highlight commuter personas that are already familiar with existing mobility platforms, demonstrating a greater propensity for adopting AVs. Figure 2 presents a demographic breakdown of these users, who tend to be high-income individuals, primarily within the 25-34 age group. Additionally, these commuters frequently use e-hailing services for social trips, such as dining out, entertainment, and recreation, as well as for their return journeys home. Understanding the characteristics of these users will be essential in designing targeted policies that encourage AV adoption within this demographic. The findings of this study underscore the importance of identifying distinct commuter personas when evaluating AV adoption potential. Multimodal commuters, who frequently switch between transportation modes, demonstrate a greater likelihood of integrating AVs into their daily travel routines. Similarly, individuals with prior experience using mobility platforms, such as e-hailing services, appear more receptive to adopting AV technology. These insights emphasize the need for a differentiated approach when promoting AVs, considering demographic factors, commuting behavior, and familiarity with digital mobility solutions. Overall, this study provides a straightforward approach for analyzing existing travel decision data from San Francisco and identifies different “user personas” that can support the efforts of understanding the types of commuters and their socio-economic background and how these factors may affect their adoption of AVs. The full paper will provide the the methodological framework in greater detail, including the development of the Binary Logistic Regression model, data preprocessing, data visualization descriptive statistics. It will also present a comprehensive analysis of multimodal commuting patterns in San Francisco. Policy recommendations will be outlined based on empirical findings, offering a roadmap for local authorities and transportation planners to facilitate a smooth transition toward AV-integrated urban mobility. Furthermore, implications for urban design, infrastructure investment, and public engagement strategies will be discussed to ensure that AVs contribute to an inclusive and sustainable transportation ecosystem.

ABSTRACT. Introduction

Urban Air Mobility promises to reshape the way aviation contributes to mobility (Pon-Prats et al., 2022). New technologies are designed to transport both passengers and goods within and across urban areas (Straubinger et al., 2020). Aerial vehicle services could be materialized by short distance vertical take-off and landing aircrafts for last-mile deliveries of freight and passenger transport. Several aircraft manufacturers and mobility companies are involved in the preparation of the air mobility advent, among them Airbus, Boeing and Uber. A list of manufacturers that have developed relevant prototypes of unmanned aerial vehicles (UAVs), that can be used for both freight and passenger transport, and those who have designed aircraft concepts is provided in the report of MITRE Corporation (2018). The developed UAV prototypes may have full or semi-automated systems for their connection to land (ie. take-off and landing operations), their capacity ranges from 2 to 4 passengers (Volocopter, 2020) and the trip length they are foreseen to cover ranges from to 35km (Volocopter, 2020) to 81km (Boeing, 2020). Hence, this new means of transport offers a broad range of new possibilities, enabling the transport of cargo, delivery services, people and even emergencies (Cohen & Shaheen, 2021). UAM is expected to contribute to the provision of more safe, sustainable and inclusive services. Insights and recommendations for the planning of UAM systems are provided by Straunbinger (2020) who reviews the requirements of UAM implementation, ranging from regulations to vehicle standards, and Rajandran and Srinivas (2020) who elaborate and discuss the arising issues of air services in UAM in the three decision-making levels (strategic, tactical, operational). UAM is envisioned to be included in SUMPs (EIP-SCC, 2018). Considering the impact that UAM could have on everyone's lives, societal acceptance is a key element if this new technology is to succeed and have a future (Pon-Prats et al., 2022). Despite numerous researchers on the subject, the topic of social acceptance doesn't seem to assume much prominence. Further attention should therefore be paid to this topic, with a focus on assessing people's perceptions of social acceptance (Rothfeld et al., 2020). The present research aims to work in this direction, considering the scarce studies that have assessed people's perceptions of social acceptance regarding UAM. Data were collected in Lisbon Metropolitan Area and a regression model was estimated to explain the aspects that affect the acceptance level of UAM services.

Literature Review

People´s perceptions on the system´s characteristics can be crucial in the diffusion process of UAVs in transport systems. During their first years of operation, safety aspects could appear as main barriers in UAV adoption (Al Haddad et al., 2020; Eker et al., 2020; Yavuz, 2024) and concerns over the potential arising risks to the public need to be assessed (Keller et al., 2018). On the other hand, positive perceptions such as the consideration of UAVs as beneficial for the society (Keller et al., 2018) and their contribution to improve transport accessibility, especially in suburban areas (Holder and Goel, 2016) can be favourable aspects for the diffusion of UAVs to the market. Psychological aspects of the potential users are very likely to be equally or more important that technological aspects (Shariff et al., 2017) and, at this point, the feelings emerging by a new technology can be determinant in the decision-making process of adopting autonomous mobility services (Schwarz, 2000;). The emotions of potential UAV users have previously been addressed by Winter et al. (2020) who through the analysis of facial expressions in an adoption regression model concluded that familiarity, value, fun factor, wariness of new technology, fear and happiness with UAVs are explanatory aspects of people´s willingness to fly with UAVs. Previous research has also indicated that socioeconomic aspects affect the intention to adopt UAVs as well. Women demonstrate lower willingness to be early adopters due to their higher safety concerns and lower trust in automation (Al Haddad et al., 2020; Coppola et al., 2024). On the contrary, the impact of higher income and young age seem to positively influence adoption (Eker et al., 2020). As sustainability considerations are increasing, passengers´ mode choice is affected favouring more environmentally friendly alternatives (Tran et al., 2020). In the era of UAM, noise can be one of the main public’s concerns for the environment. Al Haddad et al. (2020) found that noise and visual considerations of people as well as the willingness to spend a bit more for more environmentally friendly products might affect the adoption time of UAVs. Urban planning aspects is another dimension in the analysis of UAM acceptance and adoption. The residence area has contributed in previous studies to the explanation of the adoption of autonomous vehicles (Saeed et al., 2020) and UAVs (Al Haddad et al., 2020). Living in urban and suburban areas favours the diffusion of the technology.

Methodology The analysis of citizen´s perceptions at this stage of Urban Air Mobility roadmap is very important to understand the aspects that will make people accept the inclusion of this mode in their daily mobility options. It is often claimed that cost and travel time, as traditionally having been used in transport planning, drive the choices of people. This assumption holds true in many cases, however, at this stage that the inclusion of a new mode is still being explored, there is the opportunity to follow a holistic approach and analyse not only its business benefit for the operators of the system but also the added value it can bring (or not) to the potential users and the society as a whole. In an era that transport systems become complex, mobility choices increase and users´ expectations become stricter, the exploration of more aspects that in the long term will determine the adoption of a new transport mode is important. Under the assumption that industry stakeholders, policy-makers and mobility operators are designing Urban Air Mobility at a strategic level of decision-making, this study focuses on the exploration of aspects that assess the acceptance of UAVs as modes to be used on every-day mobility and get integrated in the transport systems of the cities. For this study, a survey is designed to collect information on citizens´ perceptions on acceptance and the possible time horizon of adoption. The survey design was first pilot tested to 35 respondents, related and not related to the mobility sector, who apart from their replies provided feedback for improvements in the statement structure and question flow. The survey is composed of 391 complete and valid responses from the Lisbon Metropolitan Area. Then the data was analysed with proper statistical hypothesis tests, the replies to the survey´s statements were grouped through principal component analysis and finally, a regression model for UAM acceptance was developed. 80% of the sample was used for calibration purposes and 20% for validation.

Results and Conclusion First a Principal Component Analysis was conducted to group the statements into fewer variables. Five components were designated: social transport benefits, personal transport benefits, environmental consciousness, safety concerns and pollution concerns. All the components together explained 67% of the total variance of the dependent variable and each of them individually had an Alpha-Cronbach value higher than 0.7. It was found that the higher the perceived personal and social benefits, the higher the acceptance of UAM services. The contrary was found for the level of environmental consciousness and the pollution concerns, the higher their value, the lower the acceptance. It was also found that people living in urban environments have a positive opinion of UAM services while people who need to perform short commuting trips have a negative view on UAM services. The validation results with 20% of the sample confirmed the reliability of these results. Since the roadmap of UAM implementation is still at a strategic planning phase, the results of this study can provide a starting point for the planning of the UAM ecosystem. To successfully implement new transport services, it is essential during the transition phase of exploring the possibilities of implementation to analyse the view of citizens as they are the final potential users and any change in transport systems should improve their mobility experiences.

References Al Haddad, C., Chaniotakis, E., Straubinger, A., Plotner, K. and C. Antoniou (2020) Factors affecting the adoption and use of urban air mobility. Transportation Research Part A: Policy and Practice 132, 696–712. Boeing (2020). https://boeing.mediaroom.com/2019-01-23-Boeing-Autonomous-Passenger-Air-Vehicle-Completes-First-Flight (Last accessed 20/10/2020) Coppola, P., Silvestri, F. & F. De Fabiis (2024). Heterogeneity in users’ intention-to-use Urban Air Mobility services Transportation Research Procedia, Volume 78, Pages 460-466. https://doi.org/10.1016/j.trpro.2024.02.058 EIP-SCC (2018). Urban Air Mobility initiative. https://ec.europa.eu/transport/media/news/news/2018-05-30-commission-welcomes-european-cities-joining-urban-air-mobility-initiative_en Eker, U., Fountas, G. and P. Anastasopoulos (2020) An exploratory empirical analysis of willingness to pay for and use flying cars. Aerospace Science and Technology, 104, 105993. Keller, J., Adjekum, D. K., Alabi, B. N. and B. Kozak (2018) Measuring Public Utilization Perception Potential of Unmanned Aircraft Systems. International Journal of Aviation, Aeronautics, and Aerospace, 5(3). https://doi.org/10.15394/ijaaa.2018.1243 Rajendran, S. and J. Zack (2019). Insights on strategic air taxi network infrastructure locations using an iterative constrained clustering approach. Transportation Research Part E: Logistics and Transportation Review, 128, 470–505. Straubinger, A., Rothfeld, R., Shamiyeh, M., Buchter, K-D., Kaiser, J. and K. Plotner (2020) An overview of current research and developments in urban air mobility –Setting the scene for UAM introduction. Journal of Air Transport Management, 87, 101852. Saeed, T.U., Burris, M., Labi, S. and K.C. Shina (2020). An empirical discourse on forecasting the use of autonomous vehicles using consumers’ preference. Technological Forecasting and Social Change, 158, 120130. Schwarz, N. (2000) Emotion, cognition, and decision making. Cognition and Emotion, 14 (4), 433–440. Shariff, A., Bonnefon, J.F. and I. Rahwan (2017) Psychological roadblocks to the adoption of self-driving vehicles. Nature Human Behaviour 1 (10), 694–696. Winter, S.R., Rice, S. and T.L. Lamb (2020) A prediction model of Consumer’s willingness to fly in autonomous air taxis. Journal of Air Transport Management, 89, 101926. Yavuz, Y.C. (2024). Exploring university students’ acceptability of autonomous vehicles and urban air mobility. Journal of Air Transport Management, Volume 115, 102546. https://doi.org/10.1016/j.jairtraman.2024.102546

ABSTRACT. The proverb “Time is money” highlights the importance of minimizing wasted time, referring to activities that do not generate utility. This assumption is generally applied to travel time, particularly time spent in vehicles, which is considered intermediate consumption. Estimating the value of time (VOT) in transportation is not a novel approach, as it is commonly used to assess the time savings generated by new transport investments. Traditionally, VOT is derived from survey data, such as household travel survey or discrete choice experiments.

Our paper aligns with this literature on the monetary valuation of time, but its originality lies in the data used. Specifically, we use data from controlled economic experiments conducted in a laboratory setting -- an approach still rarely utilized in the context of travel time valuation (Krcal, 2018; Randriamaro and Cook, 2022). This method provides revealed preferences, as subjects experience the real consequences of their decisions.

By transitioning to controlled laboratory experiments, we substitute travel time for waiting time in order to require subjects to wait without access to their phones, internet, books, or other activities. From this waiting time, we estimate its value and explore relationships -- whether increasing, decreasing, concave, convex -- between the value of (waiting) time and its reductions, expressed in proportional or minute reductions, and also according to the effective waiting time. We also investigate whether a single value applies across all waiting duration or wheter varying values should be considered based on waiting time length or reductions. Additionally, our experiments raise the question of small time saving (e.g., a few seconds or less than five minutes) and their relevance in the evaluation of transport investments.

ABSTRACT. The planning and implementation of large transport infrastructure projects present a complex yet pivotal challenge in fostering regional development. Such projects are part of dynamic landscapes composed of varied stakeholders, which are imposing local investment priorities and different financing mechanisms. Above and beyond their immediate function of improving mobility, these infrastructural projects exert a profound influence on the economy of a region through changes in the conditions of employment, in income levels, and in business dynamics at both a sectoral and aggregate level. In so doing, this paper provides an inclusive methodological framework for such evaluations and proffers comparative analyses of three different types of transport infrastructure: a motorway, a tourist port, and a new international tourist airport. The study aims to provide a systematic approach for understanding and assessing the socioeconomic implications of these projects, offering valuable insights for decision-makers and stakeholders involved in regional development planning. The proposed methodology encompasses an I-O modeling approach along with key performance indicators to capture the various dimensions of economic impacts of transport infrastructure projects. The I-O model represents a sound model that easily captures the interdependencies inherent in regional economies and provides estimates of direct, indirect, and induced effects of infrastructural investments. Secondly, KPIs have been adapted according to regional development objectives and have become measures of project performances along three critical dimensions: employment generation, income generation, and sectoral business expansion. Employment effects are explored in terms of the number of jobs which are created during the construction and operational phases, and income growth will also be derived based on changes in disposable household income and regional GDP contributions. Sectoral business expansion focuses on industries closely tied to transport infrastructure, such as logistics, tourism, and retail. By adopting an ex-ante approach, the framework enables stakeholders to forecast the economic outcomes of proposed projects, facilitating evidence-based decision-making and informed resource allocation. To demonstrate the applicability of this framework, the study presents a case study analyzing three infrastructure projects in Greece: a motorway, a tourist port, and a new international airport. Each project serves distinct functions and operates within unique economic contexts, offering a basis for comparative analysis. The motorway being part of the core Greek Motorway Network and TEN-T contributes to regional development via shortening travel time and reduction of transport cost. The major impacts of the motorway are felt more in the logistic, manufacturing, and trade sectors since access to larger markets was guaranteed. In the case of the motorway, there is employment growth during the construction and operational stages; due to increased accessibility, there is economic integration. The tourist port, on the other hand, primarily supports the tourism industry by facilitating maritime transport and increasing visitor arrivals. Its seasonal nature is reflected in peak economic activity aligned with tourism demand, with significant contributions to employment and growth in hospitality, entertainment, and retail sectors. The new international airport acts like a gateway to international tourists, increasing tourism and trade, thereby luring foreign direct investment. It creates a multi-faceted set of job opportunities through its construction and operational activities. In addition, connectivity via the new airport increases the scope for export-oriented industries to develop and diversify regional economies. A comparison of results has brought out both commonalities and differences in the economic effects of these infrastructure types. All three projects make a substantial contribution toward employment and income generation, though the extent and ways in which benefits are distributed differ. Their diverse impacts demonstrate the diversified infrastructure investment strategy in order to reap the comparative advantages of each type of infrastructure toward balanced and sustainable regional development. The methodological framework proposed in this study offers practical value to policymakers, planners, and stakeholders by providing tools to assess the economic impacts of transport infrastructure projects systematically. By integrating I-O modeling and KPIs, the framework enables the evaluation of both short- and long-term outcomes, supports the establishment of benchmarks, and facilitates comparisons across similar projects. Moreover, comparative analysis underscores the importance of tailoring infrastructure investments to regional needs and objectives. While motorways provide foundational connectivity that promotes interregional integration, ports and airports offer targeted benefits to tourism and trade, respectively. Together, they highlight the complementary roles of different infrastructure types in fostering regional economic development. Finally, this study contributes to the broader understanding of the socioeconomic impacts of transport infrastructure projects, offering insights into the intricate dynamics between infrastructure investments and regional economies. The findings underline the possibility of changes in employment, income, and sectoral growth that are possible with the projects and also guide that reaching their full potential needs strategic planning. A clear and flexible evaluation framework would, therefore, be apt to equip stakeholder agents with the necessary knowledge and tools to achieve credible decisions on the optimization of outcomes of the projects. This approach would be supportive towards regional development in view of economic vitality and resilience within regional business ecosystems.

ABSTRACT. The accessibility of healthcare facilities is an important indicator for evaluating the welfare of a territory’s population. Living close to a hospital or being able to reach it easily reduces indirect costs (travel, the need to stay away from home, etc.) for patients and their families; poor accessibility, on the other hand, produces high indirect costs, sometimes limiting the possibility of receiving care. A problem of equity may emerge, considering that some segments of the population may have good availability of accessible health facilities and others may not have the same levels of accessibility. When considering how and where to invest resources in building new health facilities or expanding and improving existing ones, assessing equity is important, given that the resources come from public funds. In this paper, a model is proposed to assess the equity of the location of healthcare facilities, so that a useful tool is available to optimise the investment of resources.

ABSTRACT. This work solves a case study of a transportation business that aims to plan the maximum number of jobs requested by customers that must be performed over a week in order to minimize the cost of assigning such jobs to vehicles. In this specific problem every customer pays for a job which is composed by a set of different operations (representing loading and unloading), which in turn has associated a location and a time windows. Once a job is assigned to a vehicle, all operations must be completed by the same vehicle within the specific time window. We have addressed this problem using two approaches. The first approach addresses the problem as a whole through the definition of constraints using the Hexaly solver (a black-box optimizer), while the second approach makes a partition of the problem to make the best decision at each step using a matheuristic. To compare the pros and cons of each approach, different scenarios have been considered.

ABSTRACT. This research examines the potential Advanced Air Mobility (AAM) passenger market in Hamburg, Germany, focusing on Airport Shuttle. A four-step demand model includes trip generation, distribution, and mode choice, considering travel time, cost, and passenger preferences for door-to-door AAM services. The analysis integrates pre-carriage, flight, and post-carriage phases to evaluate market size and sensitivity to travel attributes. By modeling passenger behavior and transport networks, the study highlights AAM's future role in urban mobility, emphasizing its potential for cost-effective commuting and premium services like airport shuttles. The research also identifies business opportunities for airlines and vertiport operators by analyzing market conditions and user preferences. This approach supports strategic planning and operational development for integrating AAM into metropolitan transport systems, offering insights into pricing, route design, and service optimization.

ABSTRACT. Problem In the mobility sector, big data spans sources ranging from public transport specific tools like smart cards for automated fare collection and GPS-based automatic vehicle location tracking (AVL) to broader datasets, including digital traces from mobile phones and geocoded social media records. The widespread availability of these sources has led to the passive accumulation of an unprecedented volume of precisely geo-referenced, spatio-temporal data. These data types offer a cost-effective means of analyzing human mobility patterns with a granularity and scale previously unattainable [28]. In this regard, mobile phone data are increasingly seen as a potential data source in order to develop new urban applications to support smart city objectives. Such data are relatively cheap for mobile phone operators to collect because they are passively generated for billing and network engineering purposes [27]. They have become increasingly popular in urban analysis over the last 15–20 years, because of their ability to provide space-time information on mobility patterns [24]). However, one of the primary limitation in using detailed, individual-level data is the issue of privacy, as such data can potentially enable the identification of individuals, thereby raising significant ethical and legal concerns. For Mobile Phone Data, privacy regulations vary across countries, requiring telecommunication operators to adhere to specific national frameworks. [5] propose models that promote the privacy-conscious use of mobile phone data. These models focus on ensuring that only aggregated statistical information is made available to third parties, thereby aligning with the legal requirements for the ’anonymous use of data’.

Research question The study seeks to address two key research questions. First, it investigates the strengths and weaknesses of the primary data sources employed in the transportation and mobility sectors, focusing on their advantages and limitations. These data sources include transport-specific data, such as smart card data, GPS and AVL data, APC data, as well as external information collected outside the transportation system, such as data from smart imaging, sensors, and mobile phones. Mobile phone data, in particular, provides insight into overall mobility flows within a given area rather than focusing solely on public transport users. As mobile phone data is typically made available only in aggregated form, the second research question delves into the applications and purposes of aggregated mobile phone data within the transportation and mobility sectors, emphasizing its role in advancing analysis and decision-making in these fields.

Research design The methodology employed in this study is based on a snowball literature review approach. The analysis began with the taxonomy identified by [28], which served as a foundational framework. To refine the scope and focus on the relevant field of application, keywords such as ”smart card,” ”GPS,” ”AVL,” ”APC,” ”smart imaging,” and ”mobile phone data” were combined with ”mobility” and ”transportation” in the search process. This approach ensured a comprehensive exploration of the primary data sources used in the transportation and mobility sectors. For the second research question, the analysis specifically focused on studies using mobile phone data within the context of mobility and transportation, providing a focused examination of its applications and contributions to the field.

Findings Research question 1 The findings from the first research question provide a detailed overview of the strengths and weaknesses of various data sources used in transportation and mobility analysis, based on a comparative framework. Smart Card Data (SCD) is collected primarily by public transportation (PT) companies when passengers enter vehicles, providing information such as boarding time, stop location, transport modes and stop numbers. This data can also include socio-demographic information, such as the use of student cards. Its main strength lies in the integration with PT data, like timetables and GPS data, offering insights into public transport use. However, SCD has significant limitations, including the lack of information on the used path and its exclusive relation to public transport trips [14, 21, 12, 28]. Mobile Phone Data (MPD) is gathered by telecommunications companies and provides latitude, longitude, and time-stamped mobility information. It represents one of the largest human mobility data sources, capturing movements on a broad scale. However, it is less precise than GPS data and is generally aggregated and anonymized to protect user privacy. Variants of MPD include network-based data, app-generated data, and tracking and tracing data. MPD can be effectively integrated with land-use and socio-demographic data to enhance transportation modeling [16, 1, 25, 3]. GPS and AVL Data are collected via sensors installed on vehicles, offering highly precise spatial data about transport status. This data type is particularly valuable for modeling PT systems when paired with smart card data. Despite its strengths in spatial accuracy, its usage is constrained by privacy regulations and its application to only specific vehicles equipped with GPS sensors [4, 10, 9]. Smart Imaging, Sensors, and APC Data are obtained from devices strategically placed to capture details about vehicle types, pedestrian activity, license plates, and traffic flow. These sources provide high spatial accuracy at specific locations, making them ideal for analyzing localized traffic conditions. However, they are spatially limited, can raise privacy issues and often involve significant costs for deployment and maintenance [8, 20, 18]. Research question 2 The findings of the second research question identify five core areas of application for mobile phone data in mobility and transportation: traffic management and environmental impact, improving public transport services and user experience, planning for future development, resilience, and planning and management of Large Events. Mobile phone data serves as a critical resource for traffic management and mitigating environmental impacts in urban areas. By offering real-time insights into travel behavior, congestion patterns, and traffic flow, it enables cities to implement adaptive traffic solutions and strategies to reduce emissions and manage traffic flow [17, 23, 24]. In the realm of public transport planning, mobile phone data enhances service alignment with actual passenger demand. Unlike traditional data sources like surveys, which provide static snapshots of travel behavior, mobile phone data delivers continuous, large-scale information about when, where, and how people travel. This allows transit agencies to pinpoint highdemand corridors, peak travel times, and underutilized routes with greater precision, improving both service efficiency and user satisfaction [2, 11, 22, 6]. A significant application of mobile phone data lies in planning for future transportation infrastructure and urban development. By analyzing comprehensive mobility patterns, planners can identify emerging needs within the transport network, determine areas requiring investment, and forecast future travel demand. For instance, mobile phone data highlights urban corridors with the highest travel demand, guiding investments in roadways, bike lanes, or public transport routes. This capability is integral to the realization of smart cities, where transport networks are optimized based on population behavioral patterns [26]. Mobile phone data also supports the planning and management of large events by providing near real-time information on crowd density, travel behavior, and movement patterns. This enables event organizers, city planners, and policymakers to optimize logistics, improve public safety strategies, and manage resources efficiently. Furthermore, these insights extend beyond the event itself, helping stakeholders assess the broader socioeconomic impacts on cities and surrounding regions. Studies have highlighted the potential of mobile phone data to manage crowd movements effectively and improve safety during large events [15, 7]. Additionally, such data provides crucial insights into the globalized nature of events and their transport implications, as discussed in [13] and [19].

Contributions The contribution of the first research question lies in providing a comprehensive framework of the strengths and weaknesses of various data sources in the transportation and mobility sectors. This framework can help researchers identify gaps in current knowledge, refine methodologies, and design targeted studies to address emerging challenges in the field of transportation effectively. The second research question contributes by demonstrating the versatility of AMPD as a powerful tool for mobility analysis. The study showcases how AMPD can be used not only for broad mobility pattern analysis but also for more specific applications, such as examining event-specific dynamics, conducting demographic segmentation, and assessing temporal-spatial variability. The research highlights how AMPD delivers both high-level overviews and detailed, actionable insights, making it a practical and valuable resource for a wide range of stakeholders.

References [1] Caterina Balzotti, Andrea Bragagnini, Maya Briani, and Emiliano Cristiani. Understanding human mobility flows from aggregated mobile phone data. IFAC PapersOnLine, 51(9):25–30, 2018. [2] Michele Berlingerio, Francesco Calabrese, Giusy Di Lorenzo, Rahul Nair, Fabio Pinelli, and Marco Luca Sbodio. Allaboard: a system for exploring urban mobility and optimizing public transport using cellphone data. In Machine Learning and Knowledge Discovery in Databases: European Conference, ECML PKDD 2013, Prague, Czech Republic, September 23-27, 2013, Proceedings, Part III 13, pages 663–666. Springer, 2013. [3] Jinzhou Cao, Qingquan Li, Wei Tu, Qili Gao, Rui Cao, and Chen Zhong. Resolving urban mobility networks from individual travel graphs using massive-scale mobile phone tracking data. Cities, 110:103077, 2021. [4] Noppadol Chadil, Apirak Russameesawang, and Phongsak Keeratiwintakorn. Real-time tracking management system using gps, gprs and google earth. In 2008 5th International Conference on Electrical Engineering/Electronics, Computer, Telecommunications and Information Technology, volume 1, pages 393–396. IEEE, 2008. [5] Yves-Alexandre De Montjoye, S´ebastien Gambs, Vincent Blondel, Geoffrey Canright, Nicolas De Cordes, S´ebastien Deletaille, Kenth Engø-Monsen, Manuel Garcia-Herranz, Jake Kendall, Cameron Kerry, et al. On the privacy-conscientious use of mobile phone data. Scientific data, 5(1):1–6, 2018. [6] Pierre Deville, Catherine Linard, Samuel Martin, Marius Gilbert, Forrest R Stevens, Andrea E Gaughan, Vincent D Blondel, and Andrew J Tatem. Dynamic population mapping using mobile phone data. Proceedings of the National Academy of Sciences, 111(45):15888– 15893, 2014. [7] Anthony W Dixon, Mark Henry, and J Michael Martinez. Assessing the economic impact of sport tourists’ expenditures related to a university’s baseball season attendance. Journal of Issues in Intercollegiate Athletics, 6(1):2, 2023. [8] Jos´e A Donaire, N´uria Gal´ı, and Barbora Gulisova. Tracking visitors in crowded spaces using zenith images: Drones and time-lapse. Tourism Management Perspectives, 35:100680, 2020. [9] Brian Donovan and Daniel B Work. Using coarse gps data to quantify city-scale transportation system resilience to extreme events. arXiv preprint arXiv:1507.06011, 2015. [10] Wael El-Medany, Alauddin Al-Omary, Riyadh Al-Hakim, Sufyan Al-Irhayim, and Mustafa Nusaif. A cost effective real-time tracking system prototype using integrated gps/gprs module. In 2010 6th International Conference on Wireless and Mobile Communications, pages 521–525. IEEE, 2010. [11] Daniel Elias, Friedrich Nadler, Johannes Stehno, Jens Kr¨osche, and Manuel Lindorfer. Somobil–improving public transport planning through mobile phone data analysis. Transportation Research Procedia, 14:4478–4485, 2016. [12] Michael S Frumin. Automatic data for applied railway management: passenger demand, service quality measurement, and tactical planning on the London Overground Network. PhD thesis, Massachusetts Institute of Technology, 2010. [13] Richard Giulianotti, Gary Armstrong, Gavin Hales, and Dick Hobbs. Global sport megaevents and the politics of mobility: the case of the l ondon 2012 o lympics. The British Journal of Sociology, 66(1):118–140, 2015. [14] Daniel H¨orcher, Daniel J Graham, and Richard J Anderson. Crowding cost estimation with large scale smart card and vehicle location data. Transportation Research Part B: Methodological, 95:105–125, 2017. [15] Ashok Kumar Kanaujiya and Vineet Tiwari. Crowd management and strategies for security and surveillance during the large mass gathering events: The prayagraj kumbh mela 2019 experience. National Academy science letters, 45(3):263–273, 2022. [16] Marco Mamei and Massimo Colonna. Estimating attendance from cellular network data. International Journal of Geographical Information Science, 30(7):1281–1301, 2016. [17] MM Nyhan, Itai Kloog, R Britter, C Ratti, and Petros Koutrakis. Quantifying population exposure to air pollution using individual mobility patterns inferred from mobile phone data. Journal of exposure science & environmental epidemiology, 29(2):238–247, 2019. [18] Alessandro Olivo, Giulio Maternini, and Benedetto Barabino. Empirical study on the accuracy and precision of automatic passenger counting in european bus services. The Open Transportation Journal, 13(1), 2019. [19] Rafael HM Pereira. Transport legacy of mega-events and the redistribution of accessibility to urban destinations. Cities, 81:45–60, 2018. [20] Dominik Rosenbaum, Jens Leitloff, Franz Kurz, Oliver Meynberg, and Tanja Reize. Realtime image processing for road traffic data extraction from aerial images. 2010. [21] Catherine Seaborn, John Attanucci, and Nigel HM Wilson. Analyzing multimodal public transport journeys in london with smart card fare payment data. Transportation research record, 2121(1):55–62, 2009. [22] Michael Sinclair, Saeed Maadi, Qunshan Zhao, Jinhyun Hong, Andrea Ghermandi, and Nick Bailey. Assessing the socio-demographic representativeness of mobile phone application data. Applied Geography, 158:102997, 2023. [23] Mukta Ranjan Singha and Bichitra Kalita. Using mobile phone network for urban traffic management. International Journal of Computer Applications, 65(2), 2013. [24] John Steenbruggen, Maria Teresa Borzacchiello, Peter Nijkamp, and Henk Scholten. Mobile phone data from gsm networks for traffic parameter and urban spatial pattern assessment: a review of applications and opportunities. GeoJournal, 78:223–243, 2013. [25] Jameson L Toole, Michael Ulm, Marta C Gonz´alez, and Dietmar Bauer. Inferring land use from mobile phone activity. In Proceedings of the ACM SIGKDD international workshop on urban computing, pages 1–8, 2012. [26] Davide Tosi. Cell phone big data to compute mobility scenarios for future smart cities. International Journal of Data Science and Analytics, 4:265–284, 2017. [27] Huayong Wang, Francesco Calabrese, Giusy Di Lorenzo, and Carlo Ratti. Transportation mode inference from anonymized and aggregated mobile phone call detail records. In 13th International IEEE Conference on Intelligent Transportation Systems, pages 318–323. IEEE, 2010. [28] Khatun E Zannat and Charisma F Choudhury. Emerging big data sources for public transport planning: A systematic review on current state of art and future research directions. Journal of the Indian Institute of Science, 99(4):601–619, 2019.

ABSTRACT. The introduction of Automated Vehicles (AVs) is expected to transform traffic by improving safety and traffic operation. However, benefiting from these promises is subject to the large-scale deployment of AVs, which will take a relatively long time. Thus, a coexistence of AVs and Human-Driven Vehicles (HDVs) is foreseeable in the near future where the drivers of conventional vehicles will interact with their fellow AVs which might create a complex environment. This study utilized a driving simulator to understand the dynamics of this interaction and the behavior of human drivers when interacting with increasing Market Penetration Rates (MPR) of Level 2 AVs, ranging from 0% to 75% with an increment of 25%. A total of 160 participants were recruited to interact with AVs under four different scenarios: (1) aggressive AVs with no visual recognizability, (2) aggressive AVs with visual recognizability, (3) cautious AVs with no visual recognizability, and (4) cautious AVs with visual recognizability. The results showed that human drivers adapted their behavior by reducing their average time headway and relative speed as the MPR level of AVs increased. This was more predominant when they faced cautious AVs rather than aggressive AVs, and the recognizability of AVs had no direct impact on the behavior of human drivers but elevated the experienced stress level. The safety and comfort perceptions, together with stress and traffic density levels, mediated the impact of the presence of AVs on the behavior of human drivers. The results of this study may inform the understanding of future traffic patterns in policy development.

ABSTRACT. Telework emerged in the 1970s with the rise of Information and Communication Technologies (ICT). Initially, working from home was viewed as a way to reduce the number of trips, particularly commuting trips made during peak hours (Hamer et al., 1991; Pendyala et al., 1991; Helminen & Ristimäki, 2007). In this way it could contribute to reduce congestion levels and, therefore, reduce pollutant emissions, like greenhouse gases (GHE) emissions. Over the years, researchers have studied the relationship between telework, travel behaviour, and more indirectly its effects on pollutant emissions. The obtained results have been not as positive as initially believed and advocated. Some recent studies indicate that while telework adoption may decrease the number of commuting trips, it could also lead to an increase in non-work-related trips and total distances travelled (Zhu, 2012; Melo & de Abreu e Silva, 2017; de Abreu e Silva & Melo, 2018a; de Abreu e Silva & Melo, 2018b). Furthermore, teleworkers longer accumulated travel distances can be explained by longer commuting distances, as teleworking is often linked to suburban residential locations (Mokhtarian, 2009, Melo & de Abreu e Silva, 2017; de Abreu e Silva & Melo, 2018a; de Abreu e Silva & Melo, 2018b). Recent research also reveals the existence of trade-off effects between work and non-work trips, leading to an increase in CO2 emissions (Cerqueria et al., 2020). During the COVID-19 pandemic, with government-mandated curfews and recommended or mandatory telework adoption, travel patterns changed drastically (Salazar, 2021). During this period, a strong telework adoption and a considerable decrease in the number of trips also reduced pollutant emissions (Badia et al., 2021; Eregowda et al., 2021), traffic accidents, and congestion levels (Campisi et al., 2023). Studies focused on the pandemic and post-pandemic travel impacts show a modal shift, that is, an increase in car and soft modes use (Wöhner, 2022; Costa et al., 2022) and a reduction in public transportation (Shakibaei et al., 2021; Echaniz, 2021). These dubious relationships raise questions about the effectiveness of measures such as working from home to promote more sustainable travel patterns and reduce greenhouse gas emissions. This is largely because the majority of the current emissions from commuting is caused by car, both due to its dominant share and to the higher specific emissions, which are also exacerbated by the low occupancy rates (Noussan & Jarre, 2021). Recent work using data collected during the later stages of the pandemic showed that full-day telework reduced the number of trips, although part-day telework had a contrary effect (Kappler & de Abreu e Silva, 2024). Other recent studies indicate that an increase in teleworking reduces average air pollution and GHG emissions caused by cars (Tenailleau et al., 2021; Noussan & Jarre, 2021). Since telework frequency has remained at higher levels than before the pandemic, it is of interest to study the effects of its adoption on travel behaviour and transport externalities — namely in CO2 emissions — to verify whether teleworking can be an effective transport demand management measure and promote sustainability or whether there are rebound effects in its adoption that are contrary to these objectives. To the best of our knowledge, this work uses the most complete data collection process regarding the effects of telework since the end of the COVID-19 pandemic. It uses data from a survey applied in the Lisbon Metropolitan Area (LMA) between September 2023 and February 2024. This survey includes a 7-day travel-telework diary in which telework engagement is explicitly related to the travel diary. It also considers both full- and part-days of telework, which despite being very rare in the literature, can impact travel behaviour differently. A total of 1900 valid responses were obtained. The questionnaire began with a socioeconomic characterisation of the respondent and his/her household, covering details such as the number of owned vehicles and the characteristics of the residence (type and size). It was followed by a series of Likert scale questions assessing residential location satisfaction and residential preferences. The second section focused on teleworking practices, including questions about the respondent's occupation, workplace location, telework habits in 2019, and telework frequency. This section also included another set of Likert scale questions exploring attitudes and perceptions toward telework. The final section addressed the relationship between telework and travel. Respondents were asked about their commuting mode and duration, followed by a 7-day travel-telework diary where they reported, for each day, their telework engagement and details of any trips they made, including the purpose, transport mode, and duration. The model incorporates attitudes and perceptions about telework. It is important to highlight that the survey collected data on trip duration instead of distances traveled. So, to calculate the total amount of travel related CO2 emissions by trip, we must build an emission factor of gCO2/min. The emission factor calculation includes the average energy consumption rate of each mode of transport considering vehicle occupation rate, distance travelled (in our case, we consider the average speed by mode, geographical area within the Lisbon Metropolitan Area, and time of the day). Land use characteristics are collected via secondary datasets which include variables such as density, land use mix, public transport and private car accessibility levels. Based on the collected data a conceptual model is defined (Figure 1). This model aims to understand the influence of telework engagement on travel-related equivalent CO2 emissions. First of all, the model considers the land use characteristics of residence and workplace and how could these influence commute distance, previous and current telework practices. Both the commuting distance and the previous telework experience (whether respondents worked frequently from home in 2019) are expected to influence the current attitudes about telework. These attitudes and the commuting distance influence current telework engagement which then, together with commuting distance and land use characteristics influence CO2 emissions. To test this conceptual model and its hypothesis a Structural Equation Model (SEM) is built. Sociodemographic characteristics are used here as exogenous variables in the SEM model. The results of this study are discussed in light of its policy implications, namely the possible contributions that telework could have to reduce CO2 emissions, its environmental impacts, and its potential as a travel demand management tool. References Badia, A., Langemeyer, J., Codina, X., Gilabert, J., Guilera, N., Vidal, V., Segura, R., Vives, M., & Villalba, G. (2021). A Take-Home Message From COVID-19 on Urban Air Pollution Reduction Through Mobility Limitations and Teleworking. npj Urban Sustainability, 1(1), 35. https://doi.org/10.1038/s42949-021-00037-7 Campisi, T., Tesoriere, G., Trouva, M., Papas, T., Basbas, S. (2022). Impact of Teleworking on Travel Behaviour During the COVID-19 Era: The Case Of Sicily, Italy. Transportation Research Procedia, Vol 60, 251-258, ISSN 2352-1465. https://doi.org/10.1016/j.trpro.2021.12.033. Cerqueira, E.D.V., Motte-Baumvol, B., Chevallier, L.B., Bonin, O., 2020. Does working from home reduce CO2 emissions? An analysis of travel patterns as dictated by workplaces. Transp. Res. Part D: Transp. Environ. 83. https://doi.org/10.1016/j. trd.2020.102338. Costa, C.S., Pitombo, C.S., de Souza, F.L.U., 2022. Travel behavior before and during the COVID-19 pandemic in Brazil: mobility changes and Transport policies for a sustainable transportation system in the post-pandemic period. Sustainability (Switzerland) 14 (8). https://doi.org/10.3390/su14084573. De Abreu e Silva, J., Melo, P. (2018a). Home Telework, Travel Behavior, and Land Use Patterns: A Path Analysis of British Single-Worker Households. Journal of Transport and Land Use, 11(1), 419-441. https://dx.doi.org/10.5198/jtlu.2018.1134 De Abreu e Silva, J., Melo, P. (2018b). Does Home-Based Telework Reduces Household Total Travel? A Path Analysis Approach of One and Two Worker British Households. Journal of Transport Geography, V37, 148-162. https://doi.org/10.1016/j.jtrangeo.2018.10.009 Echaniz, E., Rodríguez, A., Cordera, R., Benavente, J., Alonso, B., Sa˜nudo, R., 2021. Behavioural changes in transport and future repercussions of the COVID-19 outbreak in Spain. Transp. Policy 111, 38–52. https://doi.org/10.1016/j. tranpol.2021.07.011. Eregowda, T., Chatterjee, P., & Pawar, D. S. (2021). Impact of lockdown associated with COVID-19 on air quality and emissions from transportation sector: Case study in selected Indian metropolitan cities. Environment Systems and Decisions, 41(3), 401–412. https://doi.org/10.1007/s10669-021-09804-4 Hamer, R., Kroes, E., Van Ooststroom, H., 1991. Teleworking in the Netherlands: an evaluation of changes in travel behaviour. Transportation 18 (4), 365–382. https:// doi.org/10.1007/BF00186565. Helminen, V., Ristimäki, M., 2007. Relationships between commuting distance, frequency and telework in Finland. J. Transp. Geogr. 15, 331–342. https://doi.org/ 10.1016/j.jtrangeo.2006.12.004. Kappler, L. B., & de Abreu e Silva, J. (2024). Telework frequency and travel behaviour during the later stages of the COVID-19 pandemic: A study across different Metropolitan Areas: Lisbon, Istanbul, and Porto Alegre. Journal of Transport Geography, 123, 104102. https://doi.org/10.1016/j.jtrangeo.2024.104102 Melo, P. C., de Abreu e Silva, J., 2017. Home Telework and Household Commuting Patterns in Great Britain. Transportation Research Part A, 103, 1–24. https://doi.org/10.1016/j.tra.2017.05.011 Mokhtarian, P. (2009). If Telecommunication is Such a Good Substitute for Travel, Why Does Congestion Continue to Get Worse? Transportation Letters, 1(1):1-17. https://doi.org/10.3328/TL.2009.01.01.1-17 Noussan, M., & Jarre, M. (2021). Assessing commuting energy and emissions savings through remote working and carpooling: Lessons from an Italian region. Energies, 14(21), 7177. https://doi.org/10.3390/en14217177 Pendyala, R.M., Goulias, K.G., Kitamura, R., 1991. Impact of telecommuting on spatial and temporal patterns of household travel. Transportation 18, 383–409. https://doi. org/10.1007/BF00186566 Salazar, N. B. (2021). Existential vs. Essential Mobilities: Insights From Before, During and After a Crisis. Mobilities, 16:1, 20-34. https://doi.org/10.1080/17450101.2020.1866320 Shakibaei, S., de Jong, G.C., Alpk¨okin, P., Rashidi, T.H., 2021. Impact of the COVID-19 pandemic on travel behavior in Istanbul: A panel data analysis. Sustain. Cities Soc. 65. https://doi.org/10.1016/j.scs.2020.102619 Tenailleau, Q. M., Tannier, C., Vuidel, G., Tissandier, P., & Bernard, N. (2021). Assessing the impact of telework-enhancing policies for reducing car emissions: Exploring calculation methods for data-missing urban areas – Example of a medium-sized European city (Besançon, France). Urban Climate, 38, 100876. https://doi.org/10.1016/j.uclim.2021.100876 Wohner, Fabienne, 2022. Work flexibly, travel less? The impact of telework and flextime on mobility behavior in Switzerland. J. Transp. Geogr. 102 (C). https://doi.org/ 10.1016/j.jtrangeo.2022.103390. Zhu, P. (2012). Are Telecommuting and Personal Travel Complements or Substitutes? The Annals of Regional Science, 48:619-639. https://doi.org/10.1007/s00168-011-0460-6

ABSTRACT. Extended abstract attached to the submission

ABSTRACT. Airport access represents a significant segment оf overall travel, affecting passenger satisfaction and the efficiency оf transport systems. Different factors are valued differently by specific segments of passengers. This paper focuses on the hypothetical scenario of choosing a car or train to access the airport in the case of a weekend trip, with the aim to identify the most influential factors. The data were collected through the online survey conducted on the European level; however, this paper analyses only Serbian respondents. Results indicate that key factors influencing mode choice are cost, reliability, and demographic characteristics, with random forest model demonstrating the best performance.

ABSTRACT. Intercity bus travel plays a vital role in long-distance mobility, particularly in developing countries where cost-effective and sustainable transport options are essential. Despite its significance, existing research has largely overlooked passenger preferences for different seat types, such as seater and sleeper buses. This study addresses this gap by employing a two-stage modeling framework that integrates a demand generation model and a nested logit mode choice model to analyze intercity bus seat-type preferences in Northern India. Findings reveal that travel distance, socio-economic factors, and service attributes significantly influence passenger choices, with a notable preference for sleeper buses on trips exceeding 300 km. The study provides actionable insights for optimizing intercity bus services and supporting transport operators and policymakers in enhancing service efficiency and sustainability.

ABSTRACT. Urban mobility has become increasingly challenging due to rapid urbanization and population growth, leading to congestion, longer commute times, and public transport inefficiencies. The advancement of Automated Fare Collection (AFC) systems has provided access to extensive mobility data, enabling innovative solutions to enhance public transportation planning. This study leverages AFC data from Fortaleza, Brazil, to develop predictive models for forecasting passenger trips and optimizing transit system efficiency. This research focuses on predicting individual passenger trips using AFC data through machine learning techniques. Two primary tasks are addressed: i) forecasting the next bus line a user will take, and ii) estimating the volume of trips a user will perform over different time frames. These predictions facilitate personalized recommendations, optimize resource allocation, and enhance the commuter experience. A systematic literature review, following the PRISMA guidelines, identifies key approaches in public transport prediction. The study evaluates commonly used machine learning models, including Random Forest, XGBoost, and Neural Networks, using a dataset containing 39 million trip records from over 100,000 passengers collected between February 2023 and January 2024. The methodology involves multiple stages, including data preprocessing, feature engineering, model training, and evaluation. The dataset undergoes cleaning and structuring to ensure consistency and remove anomalies, while feature engineering extracts relevant attributes such as historical travel behavior, trip frequency, and time-of-day preferences. The various machine learning models that were tested are compared using accuracy, precision, recall, and F1-score for classification models, and mean absolute error (MAE) and root mean squared error (RMSE) for regression models. For the next trip prediction task, different classification algorithms were experimented with, including decision tree-based models, support vector machines, and neural networks. The results indicate that decision tree-based models, particularly XGBoost and Random Forest, outperform other classifiers due to their robustness in handling imbalanced datasets and their ability to capture complex data patterns. In the second task, trip volume prediction, regression-based models were implemented to forecast daily, weekly, and monthly travel demand for individual users. Similar to the classification task, decision tree-based models provided the highest accuracy, demonstrating superior performance over linear regression and neural networks. The ability to anticipate passenger volume can assist transit authorities in optimizing bus frequencies, minimizing overcrowding, and improving passenger comfort. A key contribution of this research is the clustering analysis of passengers based on mobility behavior. By segmenting users into groups according to travel frequency, preferred routes, and time-of-day preferences, it is demonstrated that personalized models yield higher accuracy in trip prediction. Additionally, the study provides valuable insights into travel patterns in Fortaleza, supporting data-driven decision-making for urban transportation planning. The findings highlight the potential of AFC data in addressing urban mobility challenges by leveraging machine learning for predictive analytics. Forecasting individual trips and travel volumes assists transit authorities in optimizing bus schedules, reducing delays, and minimizing service disruptions. Personalized recommendations for passengers can improve the commuter experience by suggesting optimal routes and travel times, increasing public transport utilization. The results indicate that decision tree-based models provide robust and reliable predictions, demonstrating adaptability to different urban settings and making them suitable for deployment in cities with AFC systems. This research contributes to the field by presenting an empirical framework for next-trip prediction, which can be adapted to other cities with similar AFC systems. Future work will focus on real-time trip forecasting and the integration of additional data sources, such as weather conditions and real-time vehicle tracking, to enhance predictive accuracy.

ABSTRACT. 1. Introduction This study describes the process of MaaS deployment and analyzes the results of a MaaS pilot trial in Madrid Region. While much literature is already available on the matter of MaaS adoption intention, studies that report the results of real-world MaaS deployments are still rare. In our study, MaaS was offered to the end user (participants of the pilot) as a new travel app. Combining the results of a survey conducted two years before with pilot results, the analysis highlights the differences and similarities between the stated and revealed preferences among the survey and pilot. 2. Pre-pilot survey on MaaS adoption intention In spring 2022 a survey was conducted in Madrid Region to assess the potential for MaaS deployment. The survey combined revealed and stated preferences: the respondents were asked to report their current use of travel apps/their functionalities and give their opinion on a set of potential functionalities of a future travel app called U-Move. Most importantly, they were also asked whether they would like to use this hypothetical app once it becomes available. The respondents were recruited by different means: face-to-face interviews, snowballing method, dissemination of flyers with a QR-code to the survey in different parts of the region, and mailing list of Madrid Regional Transport Authority. In total, 9,095 complete surveys were received from different parts of the region, with a significant share of respondents (approximately 56%) living in urban peripheries. The results of this survey guided the development of a new travel app YOVOY. 3. YOVOY MaaS features The deployment of a pilot travel app (YOVOY) was promoted by Madrid Regional Transport Authority, which ensured the participation of public and private transport operators in the pilot. The app integrated at the regional level all public transport operators (Metro, suburban rail, regional and urban buses) as well as Madrid e-bike sharing service (BiciMAD), one car-sharing operator (Zity), and one taxi operator (FREENOW). Once the origin and destination are introduced, the platform would offer a set of options starting from public transport by default, then walking, cycling, and, at last, private vehicle/taxi. The platform also enabled several customization functions such as: - Define specific public transport mode(s) - Set maximum number of transfers between modes - Set maximum walking distance to a public transport stop - Set maximum cycling distance to a public transport stop To ease the access of suburban residents to public transport networks, the app also combines car and public transport when the use of Park&Ride facility is deemed appropriate.

4. YOVOY MaaS pilot 4.1. Survey design To enable a detailed yet comprehensive analysis of pilot effects on the selected individuals, it was decided to elaborate two survey rounds: a baseline survey before app use and a post-pilot survey after 2 months of YOVOY use. The surveys were elaborated to serve two main objectives: - collect participants’ opinions and feedback about the new travel app - detect potential changes in travel habits during the 2-month pilot period. Following this logic, the first-round survey was designed to gather baseline data about respondents’ attitudes and experiences with travel apps, their disposition of using car and public transport, characteristics of their main trip (most frequent), their travel choices for secondary trips and frequency of use of different transport modes, as well as their socioeconomic characteristics. The second survey round intended to collect user experience with YOVOY and detect potential changes in respondents’ mode choice; therefore, it mainly contained questions regarding YOVOY use and its functionalities and questions related to mode choice preferences. 4.2. Participants selection Since the focus was placed on the shift from car to sustainable transport modes, only individuals with access to private vehicles could participate. The objective was to attract 200 participants. First contact with the potential participants was made in mid-September 2024 using a readily available list of potential participants from the previously conducted U-Move survey (respondents willing to participate in the follow-up studies could leave their email and were then contacted to try a new travel app). They received an email with a brief description of the new travel app and its functionalities, a pilot study plan (what would be asked from them as pilot participants), and a link to a small online survey. This online survey aimed to check that the respondent belongs to the pre-defined target group: lives and works in Madrid Region (especially in suburban areas), and has a car and a driving license. In total 2897 emails were sent, and 534 responses were received. Of them, 308 individuals fit in the target group and were admitted to the pilot study. 4.3. Pilot deployment Knowing that all pilot participants are car owners and drivers, a set of bonuses was offered to participants to entice them to try alternative transport modes and to equalize the ease of access to these alternatives with the ease of access to a car. These bonuses included: - 2 months of free public transport pass covering the whole Madrid Region - 2 months of free Madrid e-bike sharing service (BiciMAD) - discounts from private operators (10€ discount on car sharing (Zity) and 50% discount on two first rides for newly registered users of taxi (FREENOW)). Two survey rounds were planned to grasp the effect of travel app use properly: one before using the app and the other 2 months after. Each participant was assigned a randomly generated participant code, which the respondent had to enter at the beginning of each survey. This code remained the same throughout the pilot, enabling the creation of a panel dataset with the data from the first and second-round surveys (i.e., before and after using the app). 5. Pilot results The average completion time for the first survey was 10 minutes and 8 minutes for the second. A total of 237 responses was received in the first round survey, of them 224 complete. In the second-round survey, 156 responses were collected (145 complete). In total, 122 participants tried the app, but 11 did not fill in the complete survey. Thus, before/after panel dataset contains responses of 111 participants. The results suggest that stated adoption intention detected by a large-scale U-Move survey that aggregated 9095 respondents coincides quite well with revealed adoption intention reported by pilot participants. In U-Move survey, 80% of the participants claimed that they would like to try the new app once it becomes available. In YOVOY pilot 78% tried the new app (122 participants out of 156). Considering the stated vs. revealed use of a new travel app, Table 1 summarizes the findings of the three surveys (U-Move, YOVOY first and second rounds surveys) regarding situational use intention. For the stated use of travel apps, both U-Move and YOVOY first-round surveys show very similar results despite different sample sizes: about 26% of participants do not intend to use travel apps for their work/study trips. This is unsurprising as these are usual routes, and people are generally aware of available travel options. However, only 14% of those who tried a new travel app never used it for their most frequent trip, while most pilot participants did use a new app for their main trip. It is an interesting finding and can be explained by individual’s willingness to “check” whether an app is trustworthy by making it show travel options on an already known route to compare whether results correspond to individual’s knowledge/perceptions. Alternatively, it could indicate that car users are curious about travel alternatives on their main route. For trips to unknown places, the results are also quite consistent: most respondents (82% in U-Move survey and an astonishing 97% in YOVOY first-round survey) stated that they would use a travel app for this purpose, and the post-pilot survey revealed that 89% of the participants did so. This purpose appears to be the main motivation for travel app use among the respondents. Instead, it was surprising to know that 18% never used a travel app to consult public transport schedules. This is probably because, nowadays, real-time information allows us to build a route by public transport at any moment, so there is little need in knowing the exact schedule of a specific route because the app constantly updates this information depending on real-world conditions.

ABSTRACT. This study provides a statistical study of train delay data in the Swedish railway system. We assess the goodness of fit of common distributions---such as gamma, log-normal, and inverse Gaussian---to model train arrival times and identify delay patterns for each station and travel direction. The Kolmogorov-Smirnov (K-S) statistical test was applied to determine the best-fitting distributions for arrival times at ten stations. Preliminary findings indicate distinct behaviors across stations, with the log-normal distribution fitting 70% of stations. However, unique patterns, such as direction-specific delays, were observed in certain stations, highlighting the importance of localized analysis. Traditionally, in Sweden, train delays have been treated as uniformly distributed across the network, a simplification widely adopted in developing synthetic datasets for AI-based timetable rescheduling systems. This study disproves the uniformity assumption, demonstrating significant variability across locations and directions. By emphasizing the need for station- and direction-specific modeling, we provide key insights that contribute to creating more accurate synthetic datasets for AI-driven timetable rescheduling. These findings promote the development of data-driven systems that improve predictive modeling, operational efficiency, and overall reliability in railway networks.

ABSTRACT. Road safety in urban areas is a key challenge for sustainable mobility and quality of life. Pedestrians are vulnerable to risks from both vehicle traffic and unsafe behaviour. The use of smartphones, whether while driving or crossing the road, has introduced new forms of distraction that impair attention and increase the risk of accidents. In addition, non-compliance with traffic rules, such as speeding limits and running red lights, contributes to an unsafe urban environment. Advanced Traffic Management Systems (ATMS), including Red-Light Enforcement (RLE) and Automated Speed Enforcement (ASE) systems, aim to reduce accidents by encouraging safer driving. However, limited resources make it essential to strategically place these systems for maximum effectiveness. This study presents a data-driven approach that integrates real-world data with analytical and simulation models to (i) identify high-risk areas with a higher probability of accidents and (ii) determine the optimal placement of these systems to maximize their effectiveness. The proposed methodology will allow to evaluate traffic calming measures and their impact also in locations where they haven't yet been implemented. Using fixed sensors to collect data on vehicle speed and vehicle distance, we perform statistical analyses to quantify speed reductions and assess pedestrian safety by calculating the Collision Probability (CP) using a calibrated microsimulation model.The methodology has been tested in Catania (Italy), where recent investments have enabled the installation of these systems. First results show a reduction in vehicle speeds and a reduction in CP, demonstrating the potential of the tool to improve road safety. The approach is replicable and scalable, with significant practical implications for similar urban contexts.

ABSTRACT. In the Philippines, public transport services are predominantly operated by private entities under government-issued franchises, with minimal government intervention beyond franchise regulation by the Land Transportation Franchising and Regulatory Board (LTFRB). This market-driven approach has resulted in inefficiencies, such as on-street competition, service interruptions, and the use of polluting vehicles, highlighting the need for reform. This paper challenges the prevailing view of public transport as a commercial enterprise and advocates for its recognition as a public service. Greater involvement from national and local governments, including enforcing service standards and providing subsidies, is essential to ensure equitable and reliable transport access. Through interviews with six local government units (LGUs), we developed five models illustrating potential LGU roles in public transport provision. These models are: (1) Service Contracting: The LTFRB provides funding, while LGUs monitor compliance with service standards in contracts with TSEs, without collecting revenues from passenger fares; (2) LGU-Owned Franchise: LGUs hold a franchise issued by the LTFRB to operate a fleet, which can be outsourced to third-party operators through a Public-Private Partnership (PPP); (3) Fare Collection Oversight: LGUs oversee fare collection (possibly via an automatic system), with the franchise remaining with the transport service entities (TSEs), and the LGU compensates TSEs based on service contract terms; (4) Local Economic Enterprise (LEE): LGUs establish an income-generating entity to either operate a public transport fleet or manage fare collection and contract TSEs for service provision; and (5) Public Transit Authority (PTA): LGUs create a PTA, a semi-autonomous body to manage and operate public transport services, particularly in metropolitan areas. These models were evaluated for feasibility and stakeholder support by representatives from five transport service entities (TSEs), one national agency, and four LGUs. Results reveal contrasting views: while the government favors Options 2 and 4, TSEs find Option 1 most viable but criticize Options 2-5 for creating competition and raising governance concerns. The findings underscore the need for collaborative approaches to address stakeholder misalignments and ensure sustainable public transport reform.

ABSTRACT. The spread of shared mobility systems, for example ridesharing, can be a valid alternative to the use of private modes of transport. Indeed, a high number of private vehicles generate traffic congestion and air pollution. Ridesharing provides a cheaper way to travel, reducing traffic and environmental impact. It can be defined as the shared use of private cars between a group of people with similar origins and destinations. In order to support and improve this mode of transportation, it is important to identify the reasons and variables involved in their use. To this end, we took into consideration available data from the Scottish Household Survey from 2015 to 2019. Using binary logistic regressions, it was possible to highlight the most relevant variables in the use of informal ridesharing. The results showed how the non-possession of cars was the most significant factor in the use of informal ridesharing systems.

ABSTRACT. Finding sustainable solutions to the growing urban mobility problems – such as accessibility, equity, and quality of service – while managing congestion and pollution levels has been a constant challenge. Demand Responsive Transport (DRT) has emerged as an efficient service strategy to offer a reasonable level of service with significantly lower financial burden. Despite this promise, DRT services have not seen as much success in terms of its adoption, particularly in the United Kingdom. In this study, we plan to understand and evaluate the demand potential of DRT services, with Leeds City as our case example. In doing so, we also plan to capture the influence of peoples' perceptions of existing transport infrastructure and attitudes around privacy, environmental consciousness and technology. Finally, we plan to evaluate the effects of policies aimed at improving existing public transport (and its perceptions) on the uptake of DRT services.

ABSTRACT. A seamless multimodal transportation network is a cornerstone of sustainable urban mobility transition. Through implementing incentive-based strategies, a shift towards a sustainable modal split can be achieved whilst maximising the system efficiency. This paper investigates the potential of incentives to influence travellers’ behaviours towards more sustainable mode choices to promote a seamless multimodal transportation network by developing an optimisation framework to identify the optimal allocation of incentives that balances individual travel time with system-wide efficiency, while adhering to a budget constraint, ensuring cost-effective and scalable solutions for sustainable urban mobility. To address the research gaps, a bi-level programming approach is employed due to the complexity of the optimisation problem. On the upper level, the road authority aims to maximise the social benefit, e.g., to minimise total travel time and emissions by determining the optimal incentives, whereas, on the lower level, individual travellers choose their routes by minimising their personal travel time, thereby reaching an equilibrium. By applying the proposed methodology to the study area, Jätkäsaari area, Helsinki, Finland, we anticipate a significant reduction in congestion by redistributing the traffic within different modes, with a particular focus on promoting a sustainable modal shift.

ABSTRACT. Mobility as a Service (MaaS) is an innovative mobility ecosystem that should be designed so that societal and sustainability goals can be achieved. Therefore, a sound procedure to assess the sustainability level of MaaS is of paramount importance to design a new service and evaluate the characteristics of existing MaaS options. The aim of this work is to develop and implement an evaluation framework of MaaS systems, considering various assessment criteria, i.e., equity, social inclusion, safety, environmental issues, potential service users, economic sustainability for business and the performance of the transportation system. The procedure was based on a Delphi approach and an Analytic Hierarchy Process, with the integration of Fuzzy theory. Input data was obtained from a panel of experts, representing different perspectives and needs: users, transportation service providers, local authorities, and researchers. The method was applied to a MaaS pilot case. The results highlighted the flexibility of the proposed decision-support tool, which could be used by policy makers and local authorities to design or select an effective MaaS solution, considering a holistic approach including multiple goals that MaaS should seek to achieve.

ABSTRACT. Road traffic injuries are currently estimated to be the eighth leading cause of death across all age groups globally, and are predicted to become the seventh leading cause of death by 2030. Identifying the factors that influence crash injury severity and understanding their impact are crucial for the planning and implementing highway safety improvement programs. Additionally, the EU Road Safety Policy Framework 2021–2030 significantly emphasises serious injury crashes, aiming to halve their number by 2030 (Rella Riccardi et al., 2022). Understanding the factors influencing crash severity is essential for implementing effective safety measures. However, several studies have established that crashes are rare events and that the crash and non-crash cases are extremely imbalanced (Laureshyn and Varhelyi 2018; Oh et al., 2010; Cai et al., 2020). The problem of class imbalance in road traffic accident data becomes even more critical when analyzing accident severity, especially for severe and fatal crashes (Fiorenti and Losa, 2020; Hans et al., 2024). A dataset is considered imbalanced when the response variable classes are not approximately equally distributed; one class significantly outweighs the other, making it more challenging to identify and predict the underrepresented events (He et al., 2009). This problem exists in many real-world domains, such as medical diagnosis of particular cancer, oil spill detection, network intrusion detection, fraud detection (Chawla, 2003; Guo et al., 2008), and human behaviours (Song et al., 2013). In all these cases, the minority class is often the most critical to identify accurately. For example, in medical diagnosis, patients with rare diseases such as cancer typically belong to the minority class. If a cancer patient is misclassified as healthy, they will not receive the necessary treatment, potentially leading to disease progression and severe health consequences (Cao et al., 2011). This imbalance complicates predictive modelling, as traditional machine learning algorithms tend to favour the majority class, resulting in biased predictions and poor classification of severe crashes (He et al., 2009). Imbalanced datasets significantly impact the performance of classification models, particularly in non-trivial learning problems (Ndour & Dossou-Gbété, 2012). The reasons for the poor performance of the existing classification algorithms on imbalanced data sets are: (a) They are accuracy driven, i.e., their goal is to minimise the overall error to which the minority class contributes very little; (b) They assume that there is equal distribution of data for all the classes; (c) They also assume that the errors coming from different classes have the same cost (Guo et al., 2008; Ganganwar, 2012; Kumar and Sheshadri, 2012). Given these challenges, accurately classifying minority events (e.g., severe crashes) requires specialised techniques to address class imbalance. To mitigate class imbalance issues, two primary approaches have been developed:(a) Cost-sensitive learning (At the algorithm level) and (b) Sampling technique (At the data level) (Guo et al., 2008). At the algorithmic level, solutions try to adapt existing classifier learning algorithms to strengthen learning concerning the small class. Two common methods, Boosting and Cost-sensitive learning, are used in this approach (Guo and Viktor, 2004; Maheshwari et al., S., 2011). In particular, the goal of cost-sensitive learning is to minimise the cost of misclassification. Cost-sensitive learning methods enforce emphasis on the minority class by manipulating and incorporating learning parameters such as data-space weighting class-dependent cost matrix and Receiver Operating Characteristics (ROC) threshold into conventional learning paradigms (Chawla et al., 2004). These solutions alter the original class distribution at the data level, driving the bias towards the minority or positive class. They consist of resampling the original data set, either by over-sampling the minority class or by under-sampling the majority class, until the classes are approximately equally represented (Cieslak et al., 2008; KrishnaVeni, and Sobha Rani, 2011). Synthetic data generation is a more recent and promising approach to handling imbalanced data. Techniques such as the Synthetic Minority Over-sampling Technique (SMOTE) and Adaptive Synthetic (ADASYN) sampling generate artificial minority class examples to enhance model learning (Menardi & Torelli, 2010). These methods improve generalisation and reduce overfitting risks associated with simple oversampling. However, a critical limitation of these techniques is their reliance on Euclidean distance, which is not well-suited for datasets containing both numerical and categorical variables (Velleman & Wilkinson, 1993; Murphy et al., 2024; Selman et al., 2024). Euclidean distance assumes continuous numerical features, making it problematic for categorical variables without meaningful numerical relationships. Assigning arbitrary numerical values to categories can distort similarity measures, leading to inaccurate classifications. In this paper, we introduce a resampling method for dealing with response class imbalance in supervised learning with predictor variables of any type, namely numerical, ordinal, nominal, and dichotomous variables. The proposed solution may be called Synthetic Over-sampling for Numerical and CAtegorical variables (SONCA). SONCA is a balancing algorithm that generates a new synthetic dataset, ensuring the response variable is balanced across all classes. The synthetic dataset is created so that each observation, randomly drawn from the original dataset, is replaced by another observation selected from the entire predictor matrix. The random selection is performed by assuming a probability function—either triangular or Gaussian—proportional to the inverse of the distance between the randomly drawn observation from the original dataset and all observations in the predictor matrix. Unlike other algorithms in the literature, SONCA can handle datasets containing numerical and categorical data by applying a complete disjunctive encoding of the categorical predictor matrix. Furthermore, when dealing with mixed data types, the distance between the randomly drawn observation from the original dataset and all observations in the predictor matrix is computed using normalised Euclidean distance. To evaluate the effectiveness of the SONCA algorithm for both parametric and non-parametric models, the Logit model and the CART model were used. These estimation models were employed to analyse the severity of road accidents using two different datasets: PTW-ISTAT and A56 Crashes. The PTW-ISTAT dataset consists of 228,997 observations, where the response variable, composed of two classes, is highly imbalanced. The class representing accidents with injuries accounts for 98.3%, while the class of fatal accidents constitutes only 1.7%. Similarly, the A56 Crashes dataset consists of 1,831 accidents, and the response variable is also imbalanced. The class of accidents involving injuries or only material damage has a frequency of 93.1%, whereas fatal accidents make up 6.9%. To develop models that better reflect real data while minimizing overfitting, each dataset was divided into two data sets: Training set (containing 50% of the observations) used for model estimation; and Holdout set (containing the remaining 50% of the observations) used for subsequent model validation. The two partitions were randomly selected. Additionally, the observations in the holdout set were extracted to be independent of the training cases. After the dataset was split, the training set was resampled using the SONCA algorithm to balance the response variable classes, ensuring a more equitable distribution for the learning process. Performance measures were calculated on the holdout dataset for both the models estimated using the original training set and those based on the resampled dataset. The analyzed metrics include the False Negative Rate (FNrate), False Positive Rate (FPrate), True Negative Rate (TNrate), True Positive Rate (TPrate), Precision, F-measure, G-mean, Error (Err), Accuracy (Acc), and Area Under the Curve (AUC). The results of the both models estimated on the original training dataset show a high inefficacy in recognizing the positive class. The False Negative Rate is close to 1, while the True Positive Rate is nearly zero. The Acc appears high for both models, but this value is misleading since it derives exclusively from the correct classification of the negative class. The Precision, F-measure and G-mean are practically zero, confirming the inability of the models to recognize positives. This indicates that both models tend to misclassify almost all positive cases as negative. The performance measures highlight that the models estimated using the resampled training set with SONCA significantly improve the performance of both models, as demonstrated by the increase in TPrate, G-mean and AUC, although there are some differences between the CART and the Logit. This confirms that balancing the response variable enhances the classifier's ability to accurately distinguish between different outcome categories, particularly in highly imbalanced scenarios. SONCA is a user-friendly tool for data balancing, it acts as a pre-processing phase, allowing the learning system to receive observations as if they belonged to a well-balanced data set, therefore it can be applied to any learning system. SONCA is a balancing algorithm, through which it is possible to obtain a new synthetic dataset so that the response variable is balanced for each class. Differently from other algorithms in the literature, through SONCA it is possible to treat both numerical and categorical datasets.

ABSTRACT. There is a need for a platform to evaluate planned air-rail multimodal networks. The Strategic Multimodal Evaluator, developed within the MultiModX project, enables the assessment of mobility network performance by considering various factors, such as flight schedules, rail timetables, infrastructure (access, egress, and minimum connecting times), policies, and passenger preferences. Using a graph-based approach, it computes mono- and multimodal itineraries, clusters equivalent alternatives from a passenger perspective, distributes demand across these clusters, and assigns passengers to itineraries respecting vehicle capacities. Finally, it computes key performance indicators such as travel time, connectivity, and network efficiency. The Evaluator serves as a what-if analysis tool for strategic multimodal network assessment. A case study on intra-Spain mobility will demonstrate its capability to analyze policies like short-haul flight bans and their effects on regional accessibility.

ABSTRACT. This study examines human driver behavior in critical scenarios involving pedestrians to enhance our understanding of human factors in driving towards developing a credible safety baseline for automated vehicles. We examined driver-pedestrian interactions under three main scenarios: pedestrian crossing without obstruction, pedestrian crossing with obstruction but in clear sight, and pedestrian crossing with obstruction without clear sight. We compare the interactions by measuring physiological responses, driving parameters such as speed and braking position, and subjective ratings. The analysis offers initial results of the investigated scenarios, that can then be integrated into safety assessments for automated vehicles.

ABSTRACT. The real-time Railway Traffic Management Problem (rtRTMP) is key for optimizing train schedules during perturbations, aiming to minimize delay propagation and enhance the overall reliability of the railway network. However, solving the rtRTMP becomes computationally challenging when trains have many alternative routes. 

To address this challenge, the Train Route Selection Problem (TRSP) was introduced as a preprocessing step to reduce the solution space by limiting the set of alternative routes considered for each train. This study evaluates the impact of applying the TRSP in a closed-loop framework that integrates railway traffic simulation with a decision-making module, which optimizes train schedules and routes in real-time using the state-of-the-art RECIFE-MILP solver. We explore two TRSP application strategies within this framework: tactical and operational. This is the first study to experimentally validate the use of the TRSP in a realistic deployment setting where traffic plans are updated in real-time. Additionally, we analyze how closed-loop meta-parameters influence the performance of different strategies. We perform computational experiments on the Saint-Lazare station control area in France.

ABSTRACT. Railway infrastructure in densely utilized European networks is coming under increasing stress, especially on main traffic corridors shared by passenger and freight services with different speed profiles and stopping policies. Overtaking tracks allowing to change the train order are important design features to optimize capacity by allowing to schedule more trains with different characteristics and adjusting train order in operations. This work presents a comprehensive analysis of the relationship between infrastructure layout and operational stability for the case of a double-track railway line investigating the effect of additional overtaking tracks on the performance of rail corridors. A timetable ensemble generation approach allowing to obtain families of feasible timetables is adopted and the operational robustness is evaluated using a mesoscopic discrete-event simulation of train operations on the corridor considering a set of typical dispatching heuristics currently applied in practice. The approach is then applied in a use case to railway corridors in Southern Germany and the robustness of the infrastructure design is evaluated as a function of the strength of perturbation.

ABSTRACT. Introduction and motivation Being recognised as the natural backbone of any multimodal integrated transport system, to increase its competitiveness and attractiveness, rail transport is expected to deliver a step change in the reliability, flexibility, and quality of services, both for passengers and freight. To increase the rail share of transport demand, while contributing to the reduction of traffic congestion and noxious air emissions, a major target is to improve the usage of rail capacity, based on the huge potential of digitalisation and automation. The capacity of a railway line, or the numbers of trains running on it, can be increased by reducing the distances between each pair of leading-follower trains, while guaranteeing the required level of safety. The conventional method currently applied to keep trains separated, i.e., to avoid collisions, is the fixed-block signalling. In such a signalling system, tracks are physically divided into segments, named block sections, which entry points are protected by conventional lineside signals. Train position is detected through physical trackside devices, and the minimum safe headway between two trains is determined based on whole block sections, which can often lead to unnecessary long, empty stretches of railway. To decrease the “waste” of capacity induced by fixed-block signalling, which is based on discrete time and space information exchange, the concept of moving block has been introduced. Moving-block signalling implements a full radio-based train separation, without the need for physical train detection devices, and reduces the headways while guaranteeing that a train can always come to a stop in case of emergency. The system is continuously fed by real-time information, then fully adaptive to changes in train parameters, which makes it possible to increase line capacity and performance. A lot of research and innovation about rail has been, and still is, carried out in the related European Joint Undertaking Shift2Rail, and its successor Europe’s Rail, which website [1] offers many valuable documents. A recent, extensive review about rail traffic management under moving-block signalling can be found in [2], while works addressing specific aspects characterising the moving-block separation mode were published well before. For instance, in [3] a novel approach to implement moving block as a part of Communications-Based Train Control (CBTC) systems based on topology mathematics is introduced. The increasing complexity of modern railway systems, particularly in train-following dynamics, needs advanced modelling techniques to ensure safety and prompt adaptability to any possible perturbation to planned train operation. In turn, train-following dynamics under moving-block implementation require precise control to maintain safe distances between trains while responding to real-time disruptions. Traditional modelling approaches often struggle to integrate the continuous dynamics of train movement with discrete events. For instance, in [4] a discrete-time model of train dynamics is proposed to understand how stochastic disturbances (e.g., weather conditions, equipment failures, or unexpected train delays) impact train traffic flow under moving-block signalling. In this work, Hybrid Petri Nets [5] are used to design and develop a hybrid model to suitably represent train dynamics when the moving-block separation mode is implemented. Such a hybrid model is capable to describe both the continuous changes in train locations, and then the relevant time-varying headways between each pair of leading/following trains, and the unexpected discrete events that result in perturbations to a given train circulation plan. The ultimate goal is always to guarantee safety while increasing the capacity of the rail line/network under study. HPN have been used in different sectors because of their advanced capabilities to model complex systems. Petri Net (PN) was first introduced by Carl Adam Petri in 1962 [6] and after that they have been widely applied in many various fields. A reference work about PNs is still [7], which explains detailed properties, analysis and application of PNs. About HPN, a reference book is definitely [5], which gives the relevant definitions and properties, but also shows that HPN are a powerful modelling framework that combines the strengths of discrete and continuous PNs. HPN were presented in [8] as a new formalism for modelling railway systems , while in [9] a model to simulate the railway stations as hybrid dynamic systems is proposed, to analyse operation performance, structural limitations and performance evaluation during the peak hours. HPN have also been applied to other transport systems, to study such aspects as traffic signalling and level crossings. In [10] a HPN model for urban traffic networks is proposed to optimize road traffic signal coordination, reduce congestion, and improve the movement of priority vehicles (public and emergency vehicles) in a real case study. Problem description and methodology The goal of this study is to model and analyze train-following dynamics using HPN to integrate continuous variables and parameters (e.g., speed, distance) with discrete events (e.g., disruptions occurring) in train-following systems. The model represents the dynamic circulation of multiple trains operating in a moving-block separation mode whose primary objective is to maintain a safe distance between each pair of leading and following trains, even in the presence of disruptions affecting the leading train, such as speed variations, emergency braking scenarios, and poorly performing radio communications. In this context, the proposed HPN model includes distance constraints, train state transitions, and control decisions based on real-time updates. The following train must respond appropriately to these disruptions while maintaining safety margins. The proposed methodology employs an HPN model containing continuous and discrete places, as well as transitions [5]. In the HPN structure, discrete places and immediate discrete transitions are represented by circles and bars, respectively, whereas continuous places and transitions are depicted by double circles and boxes, respectively. Additionally, inhibitory and test arcs are included. The core of the proposed HPN model is illustrated in Figure 1, where continuous transitions regulate the flow based on the system's dynamic conditions, while discrete transitions manage state changes triggered by critical thresholds. More specifically, the proposed model represents a train-following system in which the transitions T_i and T_(i+1) correspond to a leader and a follower trains moving along the line. The firing speed of transition T_(i+1) varies according to the marking of discrete places. Simultaneously, the real-valued marking M(P_d ) of the place P_d indicates the distance between the trains, continuously updating their positions based on real-time information about the leader train T_(i+1). The system ensures that the actual distance between the trains is maintained, while the discrete part of the net verifies that this distance does not fall below the desired safe threshold, primarily defined by the braking distance of the following train.

Figure 1: A basic conceptual HPN diagram framework where discrete-event and continuous dynamics interact with each other More in detail, the inhibitor arc is used to prevent transition T_(s,s ̅ ) from firing until safety conditions are violated. In fact, the marking of a continuous place like P_d represents the distance (e.g., 2.15 km) between the trains, and its value is continuously updated based on the dynamic behavior of the system – specifically, the firing speeds of transitions T_i and T_(i+1), i.e., the speed of the leader train (i) and the follower train (i+1). The value of P_d changes continuously. When M(P_d )≤▁d, the immediate discrete transition T_(s,s ̅ ) fires – i.e., when the distance reaches ▁d, which is the minimum admissible distance between the trains. In this context, it is worth noting that when T_(s,s ̅ ) fires, the marking of P_d remains unchanged, thanks to the double arcs connecting P_d and T_(s,s ̅ ), which have the same weight ▁d. At the same time, the token in discrete place P_s, which models the safe nominal operating condition, moves to the unsafe place P_s ̅ , representing an unsafe condition. This mechanism ensures that the safety distance is always respected. In fact, the firing speed of T_(i+1) depends on the marking of these two places and on corrective actions implemented to adjust the distance (for instance, a reduction in the speed of train T_(i+1) according to the braking procedure when P_s ̅ is marked). On the other hand, the discrete transition T_(s ̅,s) fires when M(P_d )≥¯d, which represents the current safe distance plus a margin, indicating that the trains have once again reached a safe distance. The firing of this transition moves the token from P_s ̅ back to P_s, restoring the system to a safe condition. Note that the hysteretic thresholds ▁d and ¯d (▁d<¯d) ensure that the system avoids unnecessary transitions between safe and unsafe states. The double arc between places and transitions ensures that the value of P_d does not get updated under certain conditions after a transition fires to adjust the train’s position. This prevents the system from continuously switching between conditions due to small variations in the distance. Thanks to the proposed model, simulations will be performed to visualize the dynamic interactions between trains, analyse different scenarios (including various types of disruptions) and optimize the control logic to ensure both safety and efficiency. In particular, after setting up the model, multiple scenarios will be simulated to verify that a safe (positive) distance is maintained under all conditions, where M(P_d )=0 indicates an accident. Additionally, the system's response to different disruption patterns will be analysed, and parameters will be fine-tuned to enhance overall performance. This methodology provides a structured framework for analysing and implementing train-following systems under moving-block principles, with a strong emphasis on safety, efficiency, and adaptability to operational disruptions. The integration of HPN modelling with practical simulation tools enables both theoretical validation and practical implementation assessment. This simplified yet comprehensive model serves as a conceptual framework for safe distance management in train operations, aligning with HPN formalism while remaining adaptable for further development and analysis. Expected outcome In the complete version of the paper, how a simulation tool has been used to evaluate the performance of the proposed framework, demonstrating its advantages in enhancing railway signalling and control systems, will be explained and discussed. By simulating the model, the research aimed at assessing the real-time train dynamics under normal (planned) and disrupted (unexpected) conditions. In particular, the proposed model shows how the rail system state and line congestion affect the train operation under various scenarios and disruptions. In detail, the model validates how the following train continuously adapts its behaviour according to changes in the leading train's movement. The model is expected to evaluate the performance of system in terms of train headway, safety margin, and efficiency, comparing the results with the existing following models to improve the system. We can expect the results of this research will contribute to the advancement of next-generation railway signalling systems, supporting the transition towards more adaptive and intelligent train control systems. References [1]“Shift2Rail,” Europe’s Rail. [Online]. Available: https://rail-research.europa.eu/ [2]N. D. Versluis, E. Quaglietta, R. M. P. Goverde, P. Pellegrini, and J. Rodriguez, “Real-time railway traffic management under moving-block signalling: A literature review and research agenda,” Transp. Res. Part C Emerg. Technol., vol. 158, p. 104438, Jan. 2024, doi: 10.1016/j.trc.2023.104438. [3]H. Wang, T. Tang, C. Roberts, C. Gao, L. Chen, and F. Schmid, “A novel framework for supporting the design of moving block train control system schemes,” Proc. Inst. Mech. Eng. Part F J. Rail Rapid Transit, vol. 228, no. 7, pp. 784–793, Sep. 2014, doi: 10.1177/0954409713495015. [4]Y. Li-Xing, L. Feng, G. Zi-You, and L. Ke-Ping, “Discrete-time movement model of a group of trains on a rail line with stochastic disturbance,” Chin. Phys. B, vol. 19, no. 10, p. 100510, Oct. 2010, doi: 10.1088/1674-1056/19/10/100510. [5]R. David and H. Alla, Discrete, Continuous, and Hybrid Petri Nets, 2nd ed. Berlin/Heidelberg: Springer Berlin, Heidelberg, 2010. [Online]. Available: https://link.springer.com/book/10.1007/978-3-642-10669-9 [6]C. A. Petri, PhD Thesis, Institut fûr Instrumentelle Mathematik, Bonn, 1962. [7]T. Murata, “Petri nets: Properties, analysis and applications,” Proc. IEEE, vol. 77, no. 4, pp. 541–580, Apr. 1989, doi: 10.1109/5.24143. [8]G. Decknatel and E. Schnieder, “Hybrid Petri nets as a new formalism for modelling railway systems,” Trans. Built Environ., vol. 34, 1998. [9]F. Kaakai, S. Hayat, and A. E. Moudni, “SIMULATION OF RAILWAY STATIONS BASED ON HYBRID PETRI NETS,” IFAC Proc. Vol., vol. 39, no. 5, pp. 50–55, 2006, doi: https://doi.org/10.3182/20060607-3-IT-3902.00013. [10]A. Di Febbraro, D. Giglio, and N. Sacco, “Urban traffic control structure based on hybrid Petri nets,” IEEE Trans. Intell. Transp. Syst., vol. 5, no. 4, pp. 224–237, Dec. 2004, doi: 10.1109/TITS.2004.838180.

ABSTRACT. In many regional areas where passenger transportation by rail is available, the development of activities outside the individual’s residential location following environmentally sustainable parameters could be potentiated by setting a well-organized timetable of services addressing two crucial aspects: quality of the connectivity between stations and the length of the period at the destination. i.e., the last time service to return home. In this paper a non-periodical timetable model is developed at a tactical level, in the form of a mixed linear integer programming problem, for addressing the two previous questions. The model considers all the circulations in an operational horizon (typically a day) on a regional network, where lines with many services coexist with lines with much less services. The model sets the scheduling of arrivals and departures taking into account operational aspects such as headways, capacities and transfer times between services at stations. A preprocessing phase consisting of two steps based on a simplified version of a strategic timetable (SPOT) model is used to determine realistic upper bounds on the minimum travel times using transfers at a subset of chosen transfer stations. Results will be presented for a set of medium size test areas presenting the characteristics previously described and for different levels of congestion.

ABSTRACT. Rapid urbanisation and economic growth have resulted in high demand for transportation in cities. The widespread use of private automobiles has been linked with emissions, accidents and the misuse of public space, prompting policymakers to seek car-use reduction policies. However, they are often confronted with uncertainty about the effectiveness and suitability of those policies and their impacts on citizens. This study presents a simulation-based framework to evaluate car-use reduction measures based on a case study in Munich, Germany. The mode choice decisions of a representative sample (N=1,498) were captured through a stated preference survey. The resulting discrete choice model was integrated into an agent-based simulation, calibrated and validated against the latest household travel survey, to simulate the proposed car-use reduction policies. The first results utilising a 5% population revealed that improving neighbourhood mobility, promoting active modes and imposing parking restrictions on private automobiles may be the prevailing strategies. Subsequently, the framework will be expanded with regard to the number of agents under consideration, and the effects will be analysed spatially and for different sociodemographic segments.

ABSTRACT. Peer-to-peer (P2P) car sharing offers a promising pathway to improve the environmental sustainability and equity of urban transport systems. By complementing public transportation and active mobility options, P2P car sharing has the potential to reduce congestion, lower emissions, and promote more inclusive mobility solutions. It also addresses social equity by providing affordable alternatives for individuals who cannot afford private car ownership. While P2P car sharing is well-established in northern European countries, it remains underutilized in Italy, hindered by an ambiguous regulatory framework and limited public awareness.

This study aims to explore the demand and supply-side factors that can drive the growth of P2P car sharing in Italy, with a focus on demand forecasting and modeling. We conducted an online survey targeting residents in Italy, collecting data on preferences from both potential renters and car owners, capturing not only socio-demographic characteristics but also detailed insights into respondents' current mobility habits, perceptions, and attitudes towards the risks and potentialities of P2P car sharing. The questionnaire included a set of 12 hypothetical choice tasks designed to identify the most important technical features perceived by potential renters and car owners.

To model the potential demand and supply within this two-sided market and gain a comprehensive understanding of how its uptake could promote social inclusion and enhance the equity of the transportation system in Italy, we analyzed the data using discrete choice models. This approach enabled us to capture the key factors influencing both renters' and car owners' decisions, providing valuable insights into how P2P car sharing could foster a more sustainable and inclusive mobility landscape.

Our findings reveal that for potential renters, the primary determinants of demand are price sensitivity, particularly regarding service fees, deductibles, and overall cost transparency. Renters also place significant importance on the convenience of accessing vehicles, with proximity to available cars and the ease of unlocking the car being crucial factors influencing their decisions.

For car owners, the key factors influencing their willingness to participate in P2P car sharing include the compensation for their vehicle use, which must adequately offset the opportunity cost of sharing their car. Insurance coverage is also critical in mitigating perceived risks, while the user-friendliness of the platform—especially features such as intuitive car unlocking and seamless booking management via the app—greatly affects their decision to list their vehicle.

Additionally, both renters and car owners view the platform manager's commission as an important factor, as it impacts the service’s cost-effectiveness and overall appeal. These insights suggest that a successful P2P car sharing model must strike a balance between affordability, convenience, and trust to encourage higher participation from both renters and car owners.

These insights have important implications for both policymakers and operators. Policymakers can use these findings to develop regulations that foster a supportive environment for P2P car sharing, addressing barriers such as insurance and tax challenges and ensuring equitable access. For operators, the results offer actionable guidance for optimizing platform design, including pricing strategies, insurance options, and user experience features. By aligning the interests of both supply and demand sides, operators can enhance the scalability and sustainability of P2P car sharing as a key component of sustainable urban mobility.

ABSTRACT. The increase in life expectancy has led to a growing demand for adapted transport services for people with reduced mobility. This trend highlights the urgent need to rethink and redesign transport services, both in their structure and accessibility, to enhance the quality of service for this group. Although adapted collective vehicles are available, the saturation of public transport during peak hours significantly diminishes service quality, disproportionately affecting individuals with reduced mobility. High occupancy levels make access and comfort challenging, while the limited availability of adapted collective and discretionary vehicles falls short of meeting the growing demand. This study investigates, through a case study, the potential for shared adapted vehicles to enhance the quality of services provided to individuals with reduced mobility within urban areas. The article analyses a dedicated public transportation program incorporating discretionary vehicles under a co-payment model, financed through a combination of user contributions and local government funding. To assess the feasibility of users sharing vehicles while maintaining service quality, this work uses the route optimization tool, STORK, which readily accommodates the constraints of shared rides, allowing to easily model and analyse various sharing scenarios. STORK enabled us to evaluate the impact of ridesharing on key performance indicators, such as travel time, vehicle utilization, and service coverage.

ABSTRACT. Time-based pay-per-use bike-sharing pricing strategies can act as a monetary incentive to nudge cyclists toward minimizing their travel time, potentially leading to negative consequences for road safety. This study investigates the effects of time-based fees on bike-sharing users through a bicycle simulator experiment involving 30 participants (15 females, aged 19–28). Participants rode along a path composed of roundabouts and T-intersections under two experimental conditions presented in randomized order. In the Private Bike (PB) condition, participants were instructed to ride as they typically would in the real world. In the Bike Sharing (BS) condition, participants were informed that €0.06 would be deducted from their final payment for each minute of riding. Generalized Linear Mixed Models were employed to analyse the effects of the experimental condition on route choice at roundabouts (i.e., using the dedicated crossing versus riding within the roundabout ring), riding speed, and gap-acceptance behaviour at T-intersections. Results indicate that, in the BS condition, participants were more likely to use the roundabout ring, rode faster, and accepted shorter time gaps. These findings raise concerns regarding the safety implications of pay-per-use bike-sharing pricing strategies.