ABSTRACT. Travel behaviour research has traditionally drawn upon a well-established toolkit with foundations in economics and psychology. Early efforts by a handful of analysts to introduce machine learning into this domain were mostly seen as a distraction driven by academic curiosity. However, the rapid and undeniable advancement of AI capabilities in recent years has challenged that scepticism, prompting a re-evaluation of established methodologies and even raising fears of their obsolescence. This keynote explores how, rather than being displaced, traditional travel behaviour research has undergone a renaissance, rising, like a phoenix from the ashes, by embracing and integrating these powerful new tools. The result is not a replacement but a transformation: a more robust, insightful, and interdisciplinary approach that advances both fields in tandem.

ABSTRACT. Bike-sharing systems have become an integral part of urban mobility, offering sustainable and accessible transportation solutions. However, optimizing their usage requires a comprehensive understanding of factors influencing demand. This study investigates the determinants of shared bicycle usage in London, focusing on the effects of weather, temperature, and temporal factors, including peak demand periods. Leveraging advanced machine learning techniques such as heatmaps, decision trees (DT), support vector machines (SVM), and K-nearest neighbors (KNN), the analysis identifies key variables influencing hourly and daily demand. The findings offer practical recommendations for optimizing operational strategies, enhancing user experience, and supporting data-driven decision-making in bike-sharing systems.

ABSTRACT. Abstract This study examines the demand for shared micromobility in Rome, focusing on free-floating bike and e-scooter sharing services, using a dataset that overcomes the limitations of GBFS data. Unlike GBFS, which is typically used for station-based systems and presents challenges for free-floating services (due to frequent vehicle ID rotations and difficulties in tracking trips), this dataset is provided by Roma Mobilità and includes over 9 million trips recorded between 2022 and 2023 from seven service providers. The analysis reveals how usage patterns are influenced by seasonal variations, weather conditions, and tourist flow, with a clear preference for e-scooters in warmer months and more consistent bicycle usage in milder conditions. The highest trip volumes occur in summer, with peaks in July for e-scooters and October for bicycles. Spatially, areas with restaurants, residential buildings, and public transport connections generate the most trips. By applying a Zero-Inflated Poisson regression, the study identifies key factors influencing trip generation and attraction, providing actionable insights for urban planning. These results offer valuable guidance to optimize the deployment of shared micromobility services, enhance infrastructure, and improve user experience, ultimately making shared mobility solutions more efficient and effective for users, service providers, and the community.

ABSTRACT. The electrification of bicycles has introduced new dynamics to cycling, raising questions about how the pedal assistance of electric bikes (e-bikes) influences cyclists' route choice preferences. Understanding these preferences is crucial for developing cycling infrastructure that effectively supports the growing use of e-bikes. This paper develops a route choice framework to examine e-bike users' route preferences and compare them to those of regular cyclists, using GPS data from the E-Biking in Switzerland (EBIS) project. From the recorded trajectories 64,340 cycling trips are map-matched by using a trajectory segmentation map-matching algorithm. A data-driven path identification (DDPI) approach is employed to identify the choice set including only observed routes. A Path Size Logit (PSL) model is then estimated to identify the key determinants of e-bike users' route choices, considering factors such as gradient, type of bike facility, traffic conditions, intersections, age, and seasonal variations. To determine how e-bike users' route preferences differ from those of regular cyclists, segmented models will first be estimated separately for each bike type, followed by a pooled model. The findings will provide a comprehensive understanding of how the electrification of bicycles influences cyclists' route choice behavior, offering policymakers and transport planners valuable insights to the needs of different types of cyclists.

ABSTRACT. Please see the attached extended abstract.

ABSTRACT. Buses are essential for urban transport and generally considered safe. However, accidents involving buses, particularly in urban areas, often result in severe injuries and fatalities for vulnerable road users (VRUs). This study introduces a novel method to compute a safety envelope using reachability analysis and set-volume optimization. The research examines how the size of the safety envelope is influenced by simultaneous steering and braking maneuvers, factoring in perceived risk constraints for both standing bus passengers and VRUs. It also evaluates the relationship between minimum safety distance (MSD) and unsafe distances recorded in real-world bus-VRU collision. Key findings include the impact of VRU-perceived safety risk on a critical zone, with MSD increasing by 5 cm to 4.2 m for different maneuvers and initial speeds. Passenger-perceived safety risk also significantly affects MSD, with increases ranging from 0.05–1 m at 20 km/h to 4.8–12 m at 50 km/h, depending on the lateral offset. A comparison of MSD with crash data reveals that braking distances recorded for most of the crashes, align within the critical zone. Given levels for dangerous VRUs’ and passengers’ risk, the proposed method provides the interval when automated system should engage, as the set difference between the critical zones between the dangerous and highest risk levels.

ABSTRACT. Micromobility in urban environments has constantly undergone significant changes, such as the introduction of electric scooters. This is supported by factors such as the demand for sustainable mobility solutions driven by policies, convenience, affordability, and the flexibility two-wheeled vehicles allow. However, the increase in users and the rapid integration of these solutions pose significant challenges and concerns related to road safety. On the one hand, this paper seeks to identify contributing factors for crashes with e-scooters that yielded injuries based on historical data at a national level. On the other hand, the study explores the operational performance, safety and comfort of e-scooter users by conducting a real-world experiment in specific routes selected in the city of Aveiro, Portugal. In this case, speed, acceleration, and road grade were measured, and safety-related measures were computed, such as jerk and critical safety distance. This two-sided study will allow to understand the challenges faced when interacting with other road users and the opportunities for improvement in infrastructure and regulations. The obtained results will contribute to the development of mobility performance indicators thus creating a solid platform for future legislation and infrastructure construction.

ABSTRACT. This paper investigates driving behaviour, vehicle performance, and exhaust emissions in urban shared spaces, addressing critical challenges in road transport in terms of safety and environmental impacts. The research focuses on analysing real-world driving data collected in shared space environments, particularly the interaction of motor vehicles with vulnerable road users. The case study is in Aveiro (Portugal) and includes three mini-roundabouts and one shared space zone. Data were collected using two diesel passengers and a hybrid electric vehicle, with monitoring devices, including an integrated Portable Emissions Monitoring System (PEMS) for emissions measurement, On-Board Diagnostic (OBD-II) readers for kinetic variables, global positional systems (GPS) loggers, and video recording equipment. The research aims to validate an Integrated Driving Indicator (iDI) that combines safety, volatility, and emissions variables and comprehensively assesses driving performance. This approach evaluates driver behaviour by considering normal safety-related variables and abnormal driving patterns through acceleration and its first derivate (vehicular jerk) analysis, while simultaneously measuring global (carbon dioxide – CO2) and local (nitrogen dioxides – NO2, carbon monoxide – CO, and hydrocarbon – HC) pollutants. The results will contribute to improving driver decision support systems and foster a deeper understanding of shared space dynamics.

ABSTRACT. This study proposes an Extreme Value Theory approach to examine and quantify rear-end crash risk at siganlized intersection using traffic conflict techniques. Two hour morning peak traffic data collected using UAV from signalized intersection located in urban area. Traffic conflicts were measured by the modified time to collision (MTTC) indicator while traffic volume and average stream speed of each block were employed as covariates. A 30-meter spatial segmentation and a 1-minute temporal segmentation are combined to form spatio-temporal blocks, serving as the basis for the block maxima approach in crash frequency analysis. Using the exceedance statistics theory (also known as a Block Maxima approach corresponding to Generalised Extreme Value distribution), four different GEV models were developed, including a stationary model (i.e., GEV(0,0,0)) with no covariates and three non-stationary models (i.e., GEV(1,0,0), GEV(0,1,0), and GEV(1,1,0)) with covariates added to the location parameter, scale parameter, and both parameters of the GEV distribution, respectively. The results show that the GEV(1,1,0) is the best fitted model among the four models since considering covariates significantly improves the model performance in terms of goodness of fit. Results demonstrates the effectiveness of using spatio-temporal blocks and surrogate safety measures (SSMs) to estimate crash frequency under heterogeneous traffic conditions.

ABSTRACT. While prior studies on Transit-Oriented Development (TOD) have primarily focused on individual station performance, they often neglect the network interconnectivity that influences urban mobility and accessibility. This research is aimed at assessing the performance of TOD stations, with their dual role as a hub as well as being a normal station, within Greater Jakarta Commuter Line network. This research follows complex network principles and estimates metrics such as Degree centrality and Strength centrality besides the standard measures Betweenness centrality and Closeness centrality. The findings reveal that 85% of TOD stations exhibit strong connectivity and accessibility within the complex network. It is noted that the performance of stations is predominantly influenced by their location within the urban core, regardless of their TOD/hub status, and that the TOD stations located at terminal points performed poorer. Closeness centrality values show substantial variability, with the highest rankings concentrated in city center. Incorporating passenger flow as a weighting factor identifies a limited number of centrally positioned stations as critical connectors, while peripheral stations display lower centrality. The exponential relationships between centrality metrics indicates uniform connectivity and robust network functionality. The study underscores the importance of adopting a network-wide approach in TOD evaluations and provides strategic recommendations to enhance transit planning and policy for improved efficiency and accessibility.

ABSTRACT. The European Railway Traffic Management System (ERTMS) is being developed to enhance interoperability across European rail networks, with the European Train Control System (ETCS) as its core component. To ensure the safe and efficient execution of train movements, the Advanced Protection System (APS) is also being developed to replace traditional interlocking and trackside train protection, shifting safety functions to the train itself. By enabling flexible movement requests based on real-time network conditions, APS enhances ERTMS’s capability to optimize the planning and implementation of train journeys.

As part of the ongoing development and specification of ERTMS, the Reference Command and Control System Architecture (RCA) has been established as a modular framework comprising the Schedule Management System (PAS), the Plan Execution Module (PE), and the Automatic Train Operations Execution Module (AE). However, the implementation of ERTMS faces several challenges, particularly in terms of system integration, functional clarity, and real-world validation. The absence of standardized and thoroughly tested interfaces has led to a proliferation of proprietary solutions, complicating interoperability and hindering seamless railway operations.

This study aims to address these challenges by evaluating the PE module’s specifications through a pretotypical implementation, referred to in this article as the Interface Control (IC) module. The IC module facilitates communication between PAS and APS, ensuring the execution of Operational Plans (OPs). Testing at the Railway Operational Field Darmstadt (EBD) will assess its performance, providing insights for refining system specifications and improving ERTMS integration.

ABSTRACT. We propose a new method to generate a secondary level bus transit network, which considers explicitly the interaction with an existing primary level trunk network and allows for setting the trade-off between user and operator costs. The problem is solved by an extension of an existing heuristic. The methodology is tested numerically with a real case. Conclusions regarding the influence of characteristics of the primary level network and trade-off levels between users and operators over the resulting secondary level network are drawn.

ABSTRACT. A two-stage linear programming model is proposed to significantly improve the computational efficiency of the transit network design and frequency setting problem (TNDFSP). The first stage selects infrastructure and determines the shortest path for passengers. The second stage designs the line routing, sets line frequencies, and assigns passengers to lines. The size of the solution space of the more sizable second stage is shown to be significantly reduced due to the decisions made in the previous stage. Computations times are promising but symmetry in the problem has to be further addressed to truly benefit from the reduced solution space. To evaluate whether the gain in computational efficiency outweighs the loss in global optimality, the sequential optimization steps will be solved simultaneously as well. Ultimately, the aim of the increased efficiency is to optimize larger-scale networks with a higher level of realism to better assist transport planning in practice.

ABSTRACT. Accurate understanding of surrounding vehicle behavior is essential for safe and efficient autonomous driving. This research proposes a multi-layer LSTM model for classifying vehicle types based on their trajectories. The model processes vehicle positions and actions over time, predicting vehicle class distributions. We test the model using HighD and NGSIM datasets, with results showing its ability to accurately classify vehicle types based on driving behavior.

ABSTRACT. Connected Autonomous Vehicle (CAV) is expected to have significant impacts to the transport sector, one of which is on road traffic performance. CAV is expected to increase road capacity and reduce traffic congestion as it can communicate and thus can travel closer to each other. It is highly likely that, when the CAV operates on the road, it will co-exist with Normal Vehicle (NV). The question is, to what extent CAVs can impact when they operate in mixed traffic? Past studies have used microsimulation to represent CAVs in mixed impact but is there an alternative method that can represent traffic as a group (instead of individual vehicle) and sufficiently model traffic movement in reality?

The aim of this study is therefore to answer these two questions by developing a model for CAVs in mixed traffic by using cell transmission model (CTM) approach. The model is called Multiclass FIFO Cell Transmission Model (MF-CTM). In this model, a road link is discretised into cells and the simulation duration is discretised into timesteps. The traffic is differentiated into cohorts, with each cohort containing information namely the order number it enters a cell, the proportion of CAV and the total number of vehicles within it. By using the model, we then simulate traffic flow propagation in mixed traffic conditions. The simulation result is the evolution of cell occupancy and CAV proportion in each cell, as well as link outflow for each timestep.

The MF-CTM has two contributions to offer. Firstly, this model incorporates first-in-first-out (FIFO) framework by Bar-Gera and Carey (2022) on how vehicles (traffic cohorts) within the CTM proceed from one cell to the next. Vehicle propagation in CTM will depend on whether they follow FIFO or not, which will affect the CAV proportion in each cell. FIFO feature incorporated in the model will make vehicle movement more realistic and CAV proportion in the model more accurate compared to non-FIFO conformant models. A CTM model without FIFO has a high risk of unintended overtaking which might not be possible in reality. Secondly, by using the CAV proportion in each cell, the MF-CTM dynamically adjusts the maximum flow rate and maximum cell occupancy in each timestep. The cell with higher CAV proportion will have larger value of maximum flow rate and higher cell occupancy due to lower average vehicle reaction time and reduced jam spacing.

To ensure FIFO and achieve dynamic maximum flow rate and maximum cell occupancy, the MF-CTM consists of several formulas which are integrated into the algorithm shown in Figure 1.

To illustrate the behaviour of the MF_CTM, this study simulates a single straight 20 km link. The link is discretised into 40 cells and run for 160 timesteps. Starting from timestep 0, inflows amounting at 13.33 veh/timesteps (1600 veh/hr) come into and traverse the link through to the end before exiting the link.

At timestep 46, disruption is introduced in the link which makes the maximum flow rate of a cell in the middle of the link (or cell 20) drops to only 25% of its value. The purpose is to see the queue length and dissipation time of the traffic with different CAV proportion when a disruption is present. The disruption stays for 40 timesteps during the simulation and then removed so the maximum flow rate of the disrupted cell is restored to 100%. Comparison is also made to the simulation result of non-FIFO model or the Multiclass CTM by Levin and Boyles (2016). The purpose of such comparison is to see why it is significant to have a FIFO feature in a CTM. The computer simulation demonstrates the following:

1) Higher CAV proportion in the link inflow will result in lower queue length and dissipation time. This can be seen in Figure 1 with x-axis: timestep, y-axis: cell (or location in the link), and colour indicator: cell occupancy. In the figure, we can see that as the CAV proportion in the link inflow increases from 0.5 to 1, the queue length and dissipation time become lower. Such CAV effect cannot be represented by basic CTM, as it does not recognise the presence of different vehicle types (i.e., CAV and NV) or considers all vehicles as the same type. Therefore, if we use the basic CTM instead to carry out the simulation, then the queue length and dissipation time will be longer. This is shown in Figure 3.

2) CAV proportion across the cells varies significantly, compared to a non-FIFO model which affect the travel time of CAVs and NVs on the link. This can be seen in Figure 4 (a) and (b) with x-axis: timestep, y-axis: cell, and colour indicator: CAV proportion in the cell. In the figure, we can see that the MF_CTM can maintain the variability of CAV proportion in the cells even after vehicles are mixed together in a queue, while the non-FIFO model cannot. This is due MF_CTM adherence to FIFO when working out the movement of traffic cohorts.

3) Significant fluctuation in the link outflow rate due to the presence of CAV which MF_CTM is able capture. This is due to the significant CAV proportion variation in the link outflows as the CAV proportion within each traffic cohort is maintained through the link based on FIFO. In contrast, the non-FIFO model has only minor fluctuation due to little variation of CAV proportion in the link outflow, resulting from the mixing of traffic in the queue with the absence of FIFO feature. The CAV proportion of those mixing traffic are then averaged and reaches the end of the link in each timestep, which is not realistic. This can be seen in Figure 4 with x-axis: timestep, y-axis: outflow per timestep.

With the capabilities mentioned above, MF-CTM can provide an alternative way to accurately model CAV impact in mixed traffic while also require less computational power compared to other approach e.g., microsimulation. This study, therefore, recommends further development of MF-CTM which may include formulations for network, multilane, incorporation of public transport, varying free flow speed, etc.

ABSTRACT. Shared autonomous electric vehicles (SAEV) represent a transformative shift in urban mobility, promising enhanced flexibility, sustainability. This study investigates the impact of integrating SAEVs into a taxi-hailing aggregator, focusing on their operational efficiency and environmental benefits. Based on New York City trip data from high-volume for-hire vehicle services, a simulation is conducted to model how such vehicles would operate, exploring fleet performance, CO2 emissions, and revenue while considering constraints such as charging infrastructure, fleet size, and real-time demand fluctuations in an urban setting. The simulation evaluates key metrics, including CO2 emissions, time to pickup, lost trips rate, and net revenue, comparing them with traditional driver-based taxis. The primary results confirm expectations: initially, SAEVs operate with similar efficiency to traditional taxis due to charging constraints under limited fleet size that cannot fully meet demand. However, as the fleet size expands, missed rides decrease more rapidly for SAEVs, as they operate continuously, unlike traditional taxis, which have service gaps at night.

ABSTRACT. Connected and automated vehicles (CAVs) present a promising avenue for minimising traffic-related impacts. It is becoming a reality to envision CAVs on the roads alongside conventional vehicles (CVs). Nevertheless, a broad vision has yet to be achieved and the European Commission, for example, has been gathering efforts to combine research with investment priorities for a better understanding of the advance of this technology and reality. This work focuses on exploring the synergistic behaviour that can be induced on CAVs to improve the network performance in an urban area in a mixed environment. The case study involves a broad area that includes a roundabout near a university campus, a kindergarten and a health service. Based on real data collected from conventional probe vehicles, the network and its baseline movements, along with variables such as e.g., speed, and acceleration, are simulated in PTV VISSIM. Moreover, based on literature reference, different types of CAVs operational behaviour are simulated: cautious, normal, and all-knowing at different penetration rates. A series of simulation runs provide vehicle record data, which is used to calculate or estimate other variables, e.g., vehicle-specific power, pollutant emissions, safety distances, and comfort. Results show simulations help identify the most effective CAV operational behaviours for different traffic scenarios and that optimising the driving behaviour of CAVs, such as smoother acceleration and braking, induces an influence on CVs operational behaviour that can reduce fuel consumption and pollutant emissions. Some scenarios also led to smoother traffic movement, minimising vulnerable road users' conflicts, and reduced congestion.

ABSTRACT. In train timetabling, robustness refers to the timetable’s ability to withstand disturbances and recover from delays. However, railway systems are also subject to disruptions, major unplanned events, such as train breakdowns, that significantly impact train operations and cause substantial deviations from the timetable. Even tough disruptions are less frequent than disturbances, they have a much greater impact in practice. Disruptions often require significant rescheduling efforts, including train cancellations, rerouting, and major changes to the timetable, which often involve the closure of tracks or platforms and temporary reductions in network capacity. This reduction can create bottlenecks and increase the likelihood of delays propagating through the network causing significant inconvenience to passengers and freight transport. In this work, we aim to analyze the timetable fragility and investigate how this may be exploited to ensure reliable train service in case of disruptions. To understand the real-world implications of disruptions on timetable fragility, this work presents a case study analysis of a railway line in Norway.

ABSTRACT. In order to find a solution to maintain the market share on the banned ultra-short-haul routes, an airline can consider replacing conventional aircraft with an eclectic Vertical Take-Off and Landing aircraft (eVTOL). To make a satisfactory choice (among six eVTOLs: Joby S4, Lilium Jet, Terafugia TF-2 Lift+Push, Alice, Jaunt, and Embraer Flexcraft) while dealing with multiple criteria, the Analytic Hierarchy Process (AHP) approach can be applied. The sensitivity of eVTOLs rankings concerning different pairwise comparisons of the alternatives with respect to costs is analyzed, showing the influence of cost changes on aircraft selection. The sensitivity analysis shows that the final rank of six aircraft is sensitive to cost changes. Moreover, it is revealed that Terafugia TF-2 Lift+Push, can never be chosen as the most desirable solution, Alice and Embraer Flexcraft can never be bottom-ranked, while the other three aircraft can be both the best and the worst solution.

ABSTRACT. This research seeks to explore the way in which Low Traffic Neighbourhoods (LTNs) work. We do this by considering a range of criteria, both design specific and features relating to the area, exploring their relationship with traffic changes following LTN implementation.

In doing this, we seek to move away from binary debates that ask whether LTNs “work” or not and seek to explore how we can best design and implement LTNs to maximise their impact.

Our analysis is based on data from Thomas and Aldred’s (2023) study of traffic changes in and around 46 London LTNs. The dataset consists of over 600 datapoints from 46 LTNs in 11 London boroughs, all introduced between May 2020 and May 2021. While this does not represent the full range of LTNs delivered across the UK, it does provide us with an insight from a high number of different LTNs with the most complete dataset available. We explored the relationship with the following variables:

This paper considers traffic volume change before and after implementation of LTNs on boundary roads and on internal roads. Overall traffic volumes were considered a dependent variable to consider other variables that might influence its success.

The independent variables were focused on are listed below, calculated in the following ways:

Area: Each LTN was mapped to create a figure in square metres.

Car ownership: Using 2021 census data, we mapped this onto the LTN polygons and then produced an average figure for each area.

Public Transport Provision: Using TfL PTAL (Public Transport Accessibility Level) data, we mapped this onto the LTN polygons and then produced an average figure for each area.

Land use mix: We used the GLA’s land use ward data to calculate a non-domestic use ratio recorded in each LTN. This was then averaged as a percentage figure for each area.

Exemptions: Exemptions were calculated through exploring available information on each LTN. LTNs in which there were Blue Badge or wider resident exemptions were coded as “1”, whereas all other LTNs were coded as “0”, creating a binary dataset.

Correlation between independent variables: to ensure that we did not double count variables that are strongly correlated with one another we explored the correlation between the independent variables, resulting in the removal of “Access only network” the analysis due to its high correlation with “Area”.

Using a statistical model, we calculated the relationship between the dependent variable (traffic changes) with the independent variables.

Our analysis highlights the following relationships:

Larger LTNs are more likely to experience greater levels of traffic evaporation

LTNs with a higher proportion of non-domestic land use are more likely to experience greater levels of traffic evaporation

Other design features are unlikely to influence overall traffic evaporation

All LTNs are likely to experience traffic reduction on internal roads and the variables we analysed are unlikely to impact this.

We also explored the relationship between driver deviation and the relative convenience of active travel for 4 example LTNs. The case studies did not suggest that driver deviation correlates with LTN size and that further research is required to understand its impact.

There are some limitations to this research, which include:

The geographical context limited to London, predominantly inner London

Gaps in data from some London boroughs

Inconsistent monitoring approaches making it difficult to make comparisons between LTNs

The limited quality of data collected from Automatic Traffic Counts (ATCs)

Other variables impacting the data, such as major roadworks or proximity to other schemes.

The only scheme with an exemption was South Fulham, as other exemptions were brought in at a later date.

From this analysis, we suggest the following recommendations for LTN design:

Prioritise larger LTNs and areas with a mix of land use. However, also consider the upper limit of LTN size, particularly in relation to equity.

Smaller LTNs may still be effective, especially when considered as part of a “phased approach” to delivery, or when near/adjacent other LTNs.

Consider filter position, especially in relation to path density and driver deviation

The research also highlights the following monitoring recommendations:

Monitor more than just motor vehicle traffic levels, particularly cycle and pedestrian counts, and public transport usage, to get a better understanding of modal shift.

Establish consistent monitoring windows so that LTN evaluations are consistent with one another

Consider which boundary and internal roads to monitor. For example, by using the “path density” model we can predict where motor vehicle counts will increase, and monitor these spots accordingly, whilst also undertaking wider counts across the whole LTN area.

LTN design and prioritisation needs to consider the following equity measures as well:

When thinking about LTN size and driver deviation, consider those who have no option but to use motor vehicles for shorter journeys, particularly older people, parents of young children, and disabled people.

While car ownership is unlikely to relate to mode shift, areas of low car ownership should still be prioritised, as LTNs would unlock active travel options for those who rely on it the most, whilst also removing the negative impacts of high traffic levels from those that do not contribute towards it.

Consider additional needs based criteria that may not impact mode shift, such as access to green space or deprivation, otherwise we risk exacerbating existing inequalities.

ABSTRACT. Effective transportation planning and management rely on comprehensive monitoring of traffic conditions across entire networks. Traffic conditions can be broadly categorized into recurrent patterns from regular flows and non-recurrent patterns caused by unforeseen incidents such as accidents or adverse weather. However, non-recurrent conditions are often overlooked, leading to significant inaccuracies in traffic estimation. Sparse detector coverage further complicates data acquisition, while traditional purely data-driven and model-based approaches struggle to balance accuracy and interpretability. To address these research gaps, this study proposes a novel Physics-Informed Computational Graph (PICG) approach that combines disruption-aware deep learning models with equilibrium-aware traffic assignment models for simultaneously estimating network-wide traffic speed and flow. The PICG framework explicitly models both recurrent and non-recurrent traffic conditions, enabling precise detection and estimation of irregular disruptions, thereby facilitating targeted incident management strategies. Moreover, by explicitly isolating non-recurrent variations, PICG enhances the accuracy of recurrent traffic estimation. Extensive experiments on a simulated network demonstrate that PICG outperforms benchmark methods, offering a robust solution for diverse traffic scenarios.

ABSTRACT. The growing demand for goods, particularly in urban areas, is driving the need for innovative urban freight systems that can meet the challenges and exploit the opportunities presented by technological advancements. In fact, the traditional transport system, that uses conventional vehicles for deliveries, can result in unsatisfactory service due to delays, particularly in urban areas with high congestion. Furthermore, the use of conventional vehicles is often associated with high fuel costs and negative environmental impact. Supported by the research advances in technology, the use of new types of vehicles, such as autonomous electric vehicles (AEVs) and unmanned aerial vehicles (UAVs), which can reduce delivery times, provide a more flexible service, and reduce polluting emissions, is a challenging but promising option. An alternative solution that is becoming increasingly popular is the use of an integrated system, whereby an underused public transport system, particularly at off-peak hours, is used to transport goods [1,2,3]. Combining the transportation of goods and passengers can result in a reduction of overall costs, a decrease in emissions and traffic congestion. Integrating these features could yield innovative and promising systems, in which innovative technologies and existing infrastructures can be used to pursue efficiency and effectiveness goals, enhancing urban livability. We propose a mathematical formulation of a new delivery system in which a fleet of AEVs has to serve a set of customers in the urban area. The AEVs can be recharged by a second fleet of UAVs placed on the roofs of public transport vehicles, such as buses or trams, using an innovative wireless charging technology, namely Wireless Power Transmission (WPT) [6]. The resulting problem, namely the Autonomous Electric Vehicle Routing Problem with battery Recharging by Drones using Public Transit (AE-VRP-RDPT) consists of finding the best routes for a fleet of AEVs, serving a set of customers within their time windows, in the last mile. Since the AEVs have limited battery capacity, recharging operations are allowed to avoid service disruption. In the proposed framework, we investigate the possibility of recharging the AEVs battery by using UAVs equipped with WPT technology. UAVs may charge only one AEV while it is stopped at a customer location. In particular, a UAV can start its route from the depot as well as from some public stops (tram/bus stops) located in the urban area. We mathematically represent the problem as a Mixed-Integer Linear Programming (MILP) model, considering capacity, time windows, and synchronization constraints. We solve the MILP for studying the framework and analyzing the potential of the new paradigm. With this purpose, we conduct a comparison between two frameworks: the proposed one, and a classic one with charging stations and cable-based charging technology [4,5]. The results collected are highly encouraging, and demonstrate the effectiveness of the proposed framework, in terms of reducing both routing and overall costs.

Acknowledgement This work was supported by grants awarded by “European Union – Next Generation EU” under the “PRIN 2022 - PNRR” project: COSMO, \linebreak CUP H53D23008850001 (Giusy Macrina); and under the “PRIN 2022” project: SMOTION (ID 2022EAECWJ), CUP H53D23002000006 (Francesca Guerriero), and CUP B53D23009270006 (Luigi Di Puglia Pugliese).

References

[1] Jakob Puchinger Abood Mourad and Tom Van Woensel. Integrating autonomous delivery service into a passenger transportation system. International Journal of Production Research, 59(7):2116–2139, 2021. [2] S. Choudhury, K. Solovey, M.J. Kochenderfer, and M. Pavone. Efficient large-scale multi-drone delivery using transit networks. IEEE ICRA, pages 4543–4550, 2020. [3] Hailong Huang, Andrey V. Savkin, and Chao Huang. Round trip routing for energy-efficient drone delivery based on a public transportation network. IEEE Transactions on Transportation Electrification, 6(3):1368–1376, 2020. [4] G. Macrina, L. Di Puglia Pugliese, F. Guerriero, and G. Laporte. The green mixed fleet vehicle routing problem with partial battery recharging and time windows. Computers & Operations Research, 101:183–199, 2019. [5] M. Schneider, A. Stenger, and D. Goeke. The electric vehicle routing problem with time windows and recharging stations. Transportation Science, 48(4):500– 520, 2014. [6] H. Yan, Y. Chen, and S. Yang. Uav-enabled wireless power transfer with base station charging and uav power consumption. IEEE T. Veh. Technol., 69(11):12883–12896, 2020.

ABSTRACT. The study aims to evaluate a set of hypotheses regarding factors impacting perceived bikeability. While measures of perceived accessibility exist, e.g., the Perceived Accessibility Scale, assessments on perceived cycling accessibility are limited. We hypothesize that perceived bikeability is an important determinant of cycling behaviour, and that perceived bikeability is, in addition to socio-demographics and mobility characteristics, influenced by social and institutional support. Further, we suggest that perceived bikeability can be understood as consisting of various aspects, including comfort, convenience, safety and social support. The incorporation of social factors into perceived accessibility measures and dividing perceived bikeability into more concrete constructs remain lacking.

Using data from Bike2Green project in Stockholm, and applying factor analysis, statistical testing, and structural equation modelling, the study explores the reliability of these hypotheses. Initial results from a statistical testing indicate that perceived cycling accessibility (measured through statements from the Perceived Accessibility Scale) is positively associated with bike frequency, and mobility characteristics, alongside social environment factor of peer influence. However, further examination of the data using structural equation modelling and different specifications of the perceived bikeability factor, including one focusing on the social and institutional support, remain to be conducted.

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. Buses and other public transport vehicles use more efficient space in an urban road system than do private automobiles. Offering priority to buses at signalised intersections, particularly on single-lane approaches, where exclusive bus lanes cannot be provided, would be another strategy. The implementation of Adaptive Traffic Control Systems will help remove instabilities from the traffic flow, thus bringing the traffic closer to homogeneous conditions. This study investigates the impact of BRTS priority with ATCS on urban corridors using microsimulation in VISSIM 9.0. A novel ATCS algorithm's effects on travel time, delay, and emissions are analysed, with exhaust emissions calculated using EnViVer software. The research compares BRTS priority under fixed signal control and ATCS. Fixed signal control prioritises BRTS but lacks overall efficiency, while ATCS ensures a balanced priority for BRTS and other vehicles, demonstrating its superiority in improving traffic flow and reducing emissions.

ABSTRACT. see attached file

ABSTRACT. Purpose: The purpose of the study is to investigate GHG emissions in terms of stage lengths and capacity associated with two aviation network structures such as point-to-point (P2P) and hub-and-spoke (HS) by taking the case of Norway. In this research, we aimed to compare¬ P2P network with HS network based on three scenarios. In the first scenario, we developed a hypothetical network (HS) based on the existing data from OAG (2023) traffic analyser related to the domestic flights of Norway in 2023 and constructed a demand matrix. We compared both the networks (P2P and HS under hypothetical scenario) related GHG emissions by selecting appropriate aircraft. In the second scenario, actual-traffic related demand matrix will be developed associated to the P2P and HS networks and it will be compared for P2P and HS configurations under the actual-traffic. In third scenario, P2P-HS networks under hypothetical scenario will be compared with actual-traffic scenario to find suitable network. Introduction: Globally, passenger numbers are expected to reach 5.2 billion in 2025, a 6.7% rise compared to 2024 and the number of passengers has exceeded the five billion mark for the first time in the history of aviation (IATA, 2025). It is projected that average annual demand of passengers across the world will increase at the rate of 3.8% per year resulting over 4 billion more passengers’ movements by 2043 compared to 2023 (European Commission 2021). In the same vein, a positive demand trend in the Norwegian aviation market is forecasted with the expansion of low-cost carrier which have capture the market share successfully already. Additionally, the growth of the regional airports is seen and predicted with the continues rise of the demand and access to the small cities (Statista 2024). HS and P2P are two commonly employed primary network structures by aviation sector. HS structure emerged after the deregulation of aviation markets in the end of 1070s providing a shift in the strategies and practices of network carriers with their expectation to bring more concentration of passengers at hubs to provide a positive economic effect (Brueckner and Zhang 2001) and studies also confirm this positive economic effect brought by hubs (Brueckner and Zhang 2001; Bel and Fageda 2008). It is also investigated that the use of a HS network reduces the pollutant emissions and fuel consumption of aircraft than point to point as a whole while ensuring the lowest passenger transportation cost (Tian et al. 2020). The HS network may have a good environmental impact (lower GHG emissions) as compared to P2P network due to its benefit of economies of scale i.e., use of larger aircraft to minimize the impact as per available seat kilometers in terms of emissions. However, the environmental impact on people living near the airports can be increased due higher LTOs in HS network topology adoption compared to P2P. Therefore, it is one of the major operational issues that needs to be resolved (Pels 2021) and companies needs to optimize OD connections in their route maps, choosing between HS or P2P networks to minimize GHG emissions (Bubalo and Gaggero 2021). On the other hand, P2P networks are employed by low-cost carriers such as Ryanair. They operate under different business models as compared to HS network such as utilization of the secondary airport for P2P network. By following such strategy, they can utilize aircraft at maximum level with lower turnaround time and increased operational efficiency (Doganis 2006). Moreover, fixed cost of the airline is also another aspect that comes into play to decide the network selection between P2P and HS because airlines with higher fixed cost opt for HS network and relative lower fixed cost carriers choose P2P network (Pels et al. 2000, 2001). HS uses bigger aircraft to get environmental economy of scales, however, there are many studies support the P2P network in comparison to HS network (Morrell and Lu 2000; Peeters et al. 2001; Jamin et al. 2004; O’Kelly 2012). A scenario analysis is applied to find that P2P networks are more environmentally friendly than HS (Morrell and Lu 2000). The network with highest connectivity has lowest fuel cost. Pure single assignment HS network has highest cost of fuel (O’Kelly 2012). P2P networks have smaller global impact compared to HS network (Peeters et al. 2001). P2P networks have lower consumption of fuel and connecting flights should be replaced with direct flights (Jamin et al. 2004). Methodology: The real-world data is collected from sources such as OAG (2024) e.g., schedule analyser, and traffic analyser. Traffic analyser is used to obtain data related to actual traffic of Norway. However, types of aircraft, frequency, available seats, and seating capacity of aircraft linked data wasn't available in traffic analyser. Therefore, schedule analyser is used to get this information. Furthermore, a fixed load factor is used to calculate frequency and seating availability in our analysis. The selection of the aircraft is carried out by using bottom-up approach. Distance and demand matrix is created to select appropriate aircraft because various ranges and types of aircraft are used for passenger movements in schedule analyser database. Based on nearest distance travelled and demand, best possible aircraft is selected from the data retrieved from Schedule analyser. Furthermore, the demand estimation matrix is developed for hypothetical scenario based on all the origin and destination (OD) pairs of P2P and HS of Norway. The demand matrix based on actual traffic data scenario for P2P and HS will be developed as it is created for hypothetical scenario. The selection of the aircraft is based on the weekly demand and range (km) between the OD pair. For this estimation, first of all, Euclidean distance formulation is used to measure the closest point between range and demand. Secondly, the position of the closest point in the dataset is found to identify the most suitable aircraft based on nearest demand and range of the used aircraft in the schedule analyser. The aircraft utilized related data is obtained from schedule analyser as discussed above. The calculations of the GHG emissions are based on the selection of every aircraft in the hypothetical and actual scenarios. GHG emissions (CO2, CH4, N2O) of all the selected aircraft are calculated using the methodology proposed by EEA (2019). We considered CO2, CH4, and N2O in this study because these three pollutants comprise the highest shares of GHG emissions. In 2022, CO2 was most dominant GHG in US responsible for 80% of emissions due to human activities primarily related to fossil fuel combustion. CH4 contributed to 12% and N2O made up to 6% emissions (EPA, 2022). The applied methodology is based on TIER 3 method of estimating GHG emissions. The TIER 3 method of estimating emissions, as defined by the European Environment Agency (EEA), is part of a broader framework for estimating GHG emissions in aviation sector. It is a higher level of estimation accuracy compared to TIER 1 and TIER 2 methods, incorporating detailed data. In the FC estimation, all the flight stages are included consisting of two operational phases of the flight which are land and take (LTO) and climb, cruise, descent (CCD). For LTO, ICAO’s default FC is considered for various LTO stages such as taxi out, taxi in, take off, climb, and approach. This default setting is different for different aircraft which are available in the calculator provided by (EEA, 2019). CCD phase related FC is calculated by entering the stage length (NM) in the calculator. By summing up both LTO and CCD stages’ FC, we found the total fuel consumption which is multiplied with relevant emission index. Moreover, 100 years global warming potential is considered for the estimation of GHG emissions. In the hypothetical network, it is found that there are 456 OD pairs where the demand is 0 where no direct flight is operating. In the HS network, we considered Oslo as the hub to operate to all the spokes. As shown in Table 1, total demand from Oslo is 3,872403 passengers and 3,844333 passengers towards Oslo. In the development of hypothetical HS network, we estimated 17,281,860 passengers traffic operating from and towards Oslo which is only hub in the network considered for our study. Expected Findings: HS network can perform better when passenger per available seat kilometer (ASK) will be considered as unit of measuring the GHG emissions. ASK is a unit used to measure the carrying capacity of an airline. ASKs are calculated by multiplying the number of available seats with flown distance (EPA, 2022). It is expected that higher demand will change difference of saving of FC and emissions. In the network where the demand is higher the difference between the P2P and HS network would be less in comparison to the very thin demand. It is anticipated that HS network would perform better results as environmental economies of scale compared to P2P due to higher passenger demand. However, the difference of saving will be changed with the type of aircraft used on the specific route. Table 1: Hypothetical network data of domestic demand (Norway, 2023) Demand Type Passengers Total demand (P2P network) 12,499,298 Total demand (HS network) 17,281,860 Traffic from Oslo (Norway) 3872403 Traffic to Oslo (Norway) 3844333 Total number of OD pairs in the P2P network based on data by OAG, 2023: 2070 Contributions: This research will provide following contributions: 1. The key contribution lies in the comparison of both the networks where the findings will be established after analysing all the scenarios in the support of P2P or HS network to fill the gap in the literature. Furthermore, it will also be established that which network configuration is best to what extent in the Norwegian region. This will also be clarified that if HS network is performing better than how much efficient it is environmentally as compared to P2P. As this study evaluates 2070 routes in the Norway where P2P and HS networks are operating, it is interesting to know how efficient HS or P2P network can be where the demand is very thin. 2. Through this study, we can establish the routes where first-generation electric aircraft (FGEA) can operate either as P2P or HS network facilitators. This study will contribute to the importance of FGEA which can operate up to 400 km with 19 seats (Avogadro and Redondi 2024). 3. In the literature, it is also highlighted that with the increase in the advancement of the technologies, it is expected that the environmental economy of scales (emissions per ASK) can be achieved with HS network and it can perform better (Pels 2021). This study is incorporating new and old aircraft technologies (A320neo, A320, A319, A321, ATR42, DHC8, ERJ 190, ERJ 195) for both P2P and HS network. Therefore, it would be thought-provoking to investigate the impact of modern technologies on environment compared to old ones.

References Avogadro N, Redondi R (2024) Demystifying electric aircraft’s role in aviation decarbonization: Are first-generation electric aircraft cost-effective? Transp Res Part D Transp Environ 130:104191. https://doi.org/10.1016/j.trd.2024.104191 Bel G, Fageda X (2008) Getting there fast: Globalization, intercontinental flights and location of headquarters. J Econ Geogr 8:471–495. https://doi.org/10.1093/jeg/lbn017 Brueckner JK, Zhang Y (2001) A model of scheduling in airline networks: How a hub-and-spoke system affects flight frequency, fares and welfare. J Transp Econ Policy 35:195–222 Bubalo B, Gaggero AA (2021) Flight delays in European airline networks. Res Transp Bus Manag 41:. https://doi.org/10.1016/j.rtbm.2021.100631 Doganis R (2006) The Airline Business 1st Edition EEA (2019) EMEP/EEA air pollutant emission inventory guidebook 2019. https://www.eea.europa.eu/publications/emep-eea-guidebook-2019/part-b-sectoral-guidance-chapters/1-energy/1-a-combustion/1-a-3-a-aviation/view. Accessed 20 Feb 2024 EPA USA (2022) Overview of Greenhouse Gases. In: Earth Syst. Sci. Data. https://www.epa.gov/ghgemissions/overview-greenhouse-gases. Accessed 30 Dec 2024 European Commission (2021) EU Climate Action and the European Green Deal. 2021. https://ec.europa.eu/clima/policies/eu-climate-action_en. Accessed 11 Mar 2024 Jamin S, Schäfer A, Ben-Akiva ME, Waitz IA (2004) Aviation emissions and abatement policies in the United States: A city-pair analysis. Transp Res Part D Transp Environ 9:295–317. https://doi.org/10.1016/j.trd.2004.03.001 Morrell P, Lu CHY (2000) Aircraft noise social cost and charge mechanisms - a case study of Amsterdam Airport Schiphol. Transp Res Part D Transp Environ 5:305–320. https://doi.org/10.1016/S1361-9209(99)00035-8 O’Kelly ME (2012) Fuel burn and environmental implications of airline hub networks. Transp Res Part D Transp Environ 17:555–567. https://doi.org/10.1016/j.trd.2012.06.006 OAG (2025) OAG Analyser. https://analytics.oag.com/analyser-client/modules. Accessed 15 Dec 2024 Peeters P, Rietveld P, Schipper Y (2001) Environmental impacts of hub and spoke networks in European aviation. Vrije Univ Amsterdam 1–16 Pels E (2021) Optimality of the hub-spoke system: A review of the literature, and directions for future research. Transp Policy 104:A1–A10. https://doi.org/10.1016/j.tranpol.2020.08.002 Pels E, Nijkamp P, Rietveld P (2001) Erratum to ‘A note on the optimality of airline networks.’ In: Econ. Lett. https://scholar.google.com/scholar?hl=en&as_sdt=0%2C5&q=Erratum+to+‘“A+note+on+the+optimality+of+airline+networks”’&btnG=. Accessed 15 Jan 2025 Pels E, Nijkamp P, Rietveld P (2000) A note on the optimality of airline networks. Econ Lett 69:429–434. https://doi.org/10.1016/S0165-1765(00)00317-7 Statista (2024) Flights (Norway). https://www.statista.com/outlook/mmo/shared-mobility/flights/norway#revenue Tian Y, Sun M, Wan L, Hang X (2020) Environmental Impact Analysis of Hub-and-Spoke Network Operation. Discret Dyn Nat Soc 2020:. https://doi.org/10.1155/2020/3682127

ABSTRACT. The safety-security of pedestrian pathways, both urban and non-urban, is a priority in urban planning policies and infrastructure design. Through the application of the Haddon matrix for risk analysis and management in pedestrian paths, the aim is to provide a methodology for identifying and classifying risk factors on urban and non-urban paths. The proposed methodology combines the classical expression of risk with three main categories of variables proposed in the Haddon matrix: the characteristics of pedestrians (e.g., age, gender, walking alone or in a group), infrastructural aspects, and environmental elements. For each type of risk, the analysis examines how infrastructure design and public policies can influence injury risks through specific interventions. To complement the analysis, a three-dimensional representation model based on Cartesian axes is proposed to visualize the position of risks according to the likelihood of an incident, its impact, and Haddon’s factors. The approach is proposed to improve the safety-security, effectiveness, and accessibility of pedestrian networks, integrating innovation and digitalization with an accurate risk assessment for the creation of safer and more affordable public spaces, thus contributing to sustainable mobility and promoting pedestrian paths as ecological and safe alternatives for the community.

ABSTRACT. The green transformation of the aviation sector requires significant investments from airports, particularly in hydrogen infrastructure. Understanding the financial implications of these investments across different airport sizes and market conditions is critical for stakeholders, including airport operators, airlines, regulators, and public entities. However, existing financial reporting standards do not provide sufficient guidance for airports undergoing sustainability transitions. This study addresses this gap by developing an adaptable financial modeling framework that assesses the financial impact of sustainability investments on the three dimensions of sustainability: economic, environmental, and social. The framework is validated through a comparative analysis of two scenarios: a ‘business as usual’ scenario and a sustainability-driven airport development scenario. Financial statements and key performance indicators (KPIs) are derived and analyzed to assess the model’s applicability. The results aim to support airport operators in making informed investment decisions that balance financial viability with sustainability goals.

ABSTRACT. There has been limited applications using physiological sensor data to unpick the role of experience in preferential change under experimental conditions. We design a novel virtual reality (VR)--based data collection process that allows us to collect physiological sensor data to measure the effect of experience in a controlled setting. Specifically, we ask participants to complete a number of stated preference tasks on travel mode choice based on traditional stated preference (SP) scenarios. After each SP choice, the participant `experiences' their chosen mode in Virtual Reality (VR). The participant then ‘re-evaluates’ their choices. We develop and test different versions of a sequential sampling model (decision field theory) to evaluate how to best capture the impact of experiencing the chosen travel mode. We gain insights into the participants' relative preferences towards new travel modes and how experiencing the new modes may influence their uptake when they become available.

ABSTRACT. As cities face increasing travel demand and environmental challenges, Light Electric Vehicles (LEVs) offer a promising sustainable transport solution. While previous research has explored factors influencing LEV adoption, the impact of residential and workplace relocations on transportation mode choice remains underexplored. This study investigates how changes in housing and job locations influence LEV adoption using a stated choice experiment with 1,000 respondents in the Netherlands. Participants first evaluate hypothetical housing options, followed by transport mode choices. Then, they assess job alternatives and make another transport mode choice. A mixed logit model will be estimated to identify key factors affecting transportation mode choice, with a particular focus on LEV adoption. The findings will provide insights into urban planning and policy strategies to promote LEV adoption as an alternative to car-based travel.

ABSTRACT. Walking offers significant societal, public health, and sustainability benefits, driving cities to enhance accessibility and walkability. This has spurred research into pedestrian behavior and movement, yet destination choice models remain underexplored, requiring a better understanding of environmental influences and realistic choice set generation.

Realistic choice set generation is crucial since it directly impacts model estimation outcomes. Researchers have used various methodologies—deterministic rules (de Dios Ortúzar et al., 2024) and sampling methods (Ben-Akiva & Lerman, 1985)—with most studies relying on random sampling. For instance, Clifton et al. (2016) developed a model using ten randomly selected alternatives within a three-mile (4.8 km) network distance from each origin, while Eash et al. (1999) formed sets of 50 possible destinations within a two-mile (3.2 km) X-plus-Y distance. Similarly, Khan et al. (2014) employed 40 alternatives (including the actual destination), and Berjisian et al. (2019) used both random and stratified importance sampling, yielding 112 alternatives per pedestrian at the PAZ level.

However, many approaches depend on fixed distance thresholds and alternative counts, which can lead to computational challenges and the exclusion of key destinations when aggregation levels vary. Enhancing choice set generation accuracy is therefore essential for reliable model estimation.

The literature has not thoroughly examined how different choice set generation methods affect destination choice model estimates. This study addresses that gap by proposing a novel graph-based approach and comparing it with distance-based generation using household travel survey data from Melbourne and Brisbane.

ABSTRACT. Climate change, together with the growth of urban populations and e-commerce, highlights the need for innovative and sustainable mobility solutions that provide attractive services for both people and goods. In this respect, we propose an optimization model for a mixed people/goods transport network that integrates the concepts of Physical Internet (PI), ride-sharing, and crowd-shipping into a single framework. In this unified framework, drivers offer a certain transport capacity and agree to transport passengers and/or goods along their trips. According to the PI concept, goods can be disassembled, transported using different vehicles, and reassembled at their destination. This approach allows for better utilization of available transport capacity, increases the flexibility of allocations between goods and drivers, and reduces the need for additional vehicles. A centralized transport supply manager collects all travel requests and offers, determines the best allocation between drivers and people/goods, and optimizes driver routes and schedules. The proposed model is applied to a simple case study to test its effectiveness. Results show that the model effectively allocates people and goods to drivers while ensuring that all pick-up and delivery time windows are met, the number of allocated people and goods is maximized, and driver travel times are minimized. This model contributes to the development of innovative urban transport schemes that can serve as an alternative or complement to traditional public transport and logistic systems. By increasing the efficiency of urban mobility, it aims to reduce car dependency, pollution and traffic congestion while simultaneously improving vehicle occupancy rates.

ABSTRACT. Introduction and Objectives According to the EU Green Paper on Urban Mobility, three clear challenges must be addressed: decreasing congestion, enhancing the quality of public transport services, and improving accessibility by fostering active modes. Although mobility differs significantly between developing and developed countries, the problems that must be faced are similar in both contexts but to a different extent. In most developing countries, private cars are the dominant mode of transportation for several reasons. One of them is the differential quality of public transport services and the social role of PT in allowing access to different destinations for different groups of citizens. For this analysis, it is necessary to consider contextual variables and cultural roles of different groups of users and non-users. This research investigates through quantitative and qualitative methods which are the key drivers explaining differences between two similar cities in two countries regarding mobility patterns and differences among bus users. Since buses are the only PT mode available in most developing countries, they have become the cornerstone for environmental and social targets. Buses as a massive transport alternative, ease traffic flow to reduce the congestion caused by the extensive usage of private vehicles. On the other hand, the bus system also plays a key social role providing convenient access to the main destinations regardless of their economic level, gender, residential location, etc. In this context, the objective of this research is to compare how bus services meet citizen mobility needs and social expectations through qualitative and quantitative analysis of similar long surveys done in the cities of Tangier (Marrocco) and Oviedo (Spain)

Case studies for cross-country analysis Tangier and Oviedo are case studies representing urban dynamics in developing and developed countries. Tangier is a rapidly growing city facing challenges regarding quality of services and access to PT. The bus is the primary mode of PT, whose young population depends on daily trips. In contrast, Oviedo, a medium-sized European city, has a well-established PT network. It has high user satisfaction, extensive service coverage, and integration with other mobility options; a wide range of users, including elders, and a high share of female users.

Table 1: The Local Conditions: Tangier vs Oviedo

Comparative mobility patterns among bus users The survey consisted of 21 questions divided into three sections: socioeconomic characteristics, trip characteristics, and bus service satisfaction. The bus service satisfaction section was measured using a Likert scale from 1 to 5, where 1 represents "totally unsatisfied" and 5 represents "totally satisfied". This structured approach allowed for a comparative analysis of Tangier and Oviedo's public transport performance and perceptions. Table 2: Survey comparison: Tangier vs Oviedo

The bus´ users mobility patterns show significant differences shaped by socioeconomic conditions. Survey data from both cities indicate that bus dependency is notably higher in Tangier, where 73% of respondents use buses daily, compared to 51% in Oviedo. Trip purposes differ as well: in Tangier, commuting for work (44%) and education (19%) are the primary reasons for using buses, while in Oviedo, leisure trips (30%) and commuting (36%) are more balanced. Additionally, car ownership is much lower in Tangier, with only 4% of respondents owning a private vehicle, whereas 40% of Oviedo residents have access to a car. These differences highlight the role of buses as an essential and often sole mobility option in Tangier, while in Oviedo, public transport competes with private vehicle usage. ALSA provided data of number of lines, services, timetables for both cities, which are the basis for the comparison with the passenger opinions. Also, the research group collected information in both cities about quality of installations, bus stops, information, ticketing, etc.

Table3.comparison of mobility's main differences

Comparative analysis on quality of bus services satisfaction The perception of bus service quality differs significantly between the two cities. In Oviedo, higher satisfaction levels are recorded for comfort, accessibility, and service information, reflecting a more developed transport infrastructure. However, challenges exist in frequency and availability on weekends, where improvement is needed. In contrast, Tangier's bus users express lower satisfaction levels, particularly regarding service reliability, comfort, and access to real-time information. However, ticket pricing is seen as a strong advantage in Tangier, making public transport more economically accessible.

Table 4 Assessment of users' satisfaction with the bus service attributes: Tangier vs Oviedo

Cross-country characterization of bus users’ groups through Cluster Analysis In Tangier, users are grouped into the four clusters shown in Table 5. These groups reflect a strong dependency on PT for essential activities, like work and education. In contrast, Oviedo´s clusters show a more balanced mix of work, social, and leisure-oriented travel. The presence of a significant elderly cluster highlights the social role, while Tangier’s younger population uses buses primarily for reaching destinations for work and study.

Table 5 Comparison of Bus User Clusters in Tangier and Oviedo

Results of quantitative vs qualitative analysis and policy recommendations The analysis highlights fundamental differences in mobility patterns and user groups between Tangier and Oviedo, shaped by socioeconomic and urban contexts. The quantitative analysis, based on operation data and survey responses, identified key disparities in service satisfaction, travel behavior, and demographic profiles, revealing that Tangier’s users have higher bus dependency, while Oviedo’s users prioritize service quality and flexibility. The qualitative analysis, derived from user feedback and contextual factors, highlighted cultural and economic influences shaping mobility patterns, such as Tangier’s affordability-driven reliance on buses versus Oviedo’s structured and choice-based transport usage. Tangier's users are primarily young workers and students who rely on buses for daily activities. In contrast, Oviedo's users include many retirees and occasional travelers, reflecting greater transport flexibility and alternative mobility options. Oviedo excels in comfort, accessibility, and service information, whereas Tangier faces challenges with reliability, infrastructure, and network coverage despite offering more affordable fares. These results allow us to design customized policy measures to improve services and make PT transport more attractive to reduce car dependency and improve sustainability.

ABSTRACT. Order dispatching is instrumental to online ride-hailing platforms, which continuously match rider trip requests to drivers. The order-dispatching problem is one of the typical decision-making problems under uncertainty, as parameters such as the pick-up time and the acceptance rates for each order etc. need to be determined before solving the matching optimization problem. Conventional approaches to such decision-making problems with uncertainties typically employ a two-stage process: first estimating uncertainties and subsequently solving the matching optimization problems. However, inferior upstream predictions may yield suboptimal solutions in downstream decisions. To this end, we first propose a novel circular structure for the order-dispatching problem that accounts for the interaction between platform matching via integer programming and individual choice models via regret-based models. In addition, we develop a decision-focused learning (DFL) framework for the order-dispatching problem with multiple upstream modules to accommodate multiple uncertain parameters. We also introduce the Algorithm Unrolling approach to convert the cyclic structure into a sequential one, enabling efficient gradient propagation. Numerical experiments are conducted using two real-world ride-hailing datasets. The novel DFL framework with multiple upstream parameters demonstrates significant potential for generalization across various problem contexts in intelligent transportation systems.

ABSTRACT. Please find our extended abstract enclosed as a pdf.

ABSTRACT. In light of the COVID-19 pandemic, potentially permanent changes in patterns and preferences for work and travel have persisted, which may lead to a shift in residential preferences. The latter may potentially affect residential location choice in the long-term, which may, in turn, re-shape the structure of urban areas. Therefore, the aim of this study is to investigate the determinants of relocation intentions based on empirical data collected over a year and a half after pandemic restrictions were fully lifted in Scotland, UK. It was found that future relocation intentions among the sample are predominantly connected with dwelling attributes. However, accessibility still appears to hold importance in future relocation intentions for non-commuting purposes, while the importance of workplace accessibility seems to decline. Thus, while there is a possibility that future relocation intentions may favour less busy areas, it is more likely that a compromise between the desirable traits of urban and rural/suburban areas would be preferred as opposed to an outright urban-to-rural migration trend.

ABSTRACT. The implementation of 30 km/h (20 mph) speed limits has gained global attention as a strategy to enhance road safety and urban livability. While such interventions have been widely adopted across Europe and North America, Greece is at early stages of exploring their potential benefits. This study examines public perceptions of citywide 30 km/h speed limits in Greece, assessing levels of support, anticipated safety improvements, and broader implications for urban quality of life. Data was collected through an online survey targeting urban and rural residents, yielding 718 valid responses. Statistical analysis, including ordered probit modeling, identified key factors influencing public acceptance. Findings reveal that over 50% of respondents support the intervention, with many expecting reductions in air and noise pollution, as well as improved walkability. Younger individuals, public transport users, and those in single-person households were more likely to support the speed limit, whereas daily car users exhibited opposition and skepticism towards a 30 km/h speed limit. The results suggest that while road safety remains a key motivator, respondents also acknowledge the possibility of benefits in urban livability. These insights can potentially inform policymakers in designing effective speed limit policies that balance mobility needs with sustainability goals.

ABSTRACT. Walking School Bus (WSB) has been implemented in different countries to promote children’s active travel to school and address health, safety, and social challenges associated with increasing car dependency. Existing studies on parents’ school travel behaviour, particularly related to WSB, is predominantly qualitative, focusing on the experiences of individual parents who have used the WSB. However, there is a notable lack of research investigating parents’ mode choice behaviour or more precisely intention to switch to this new mode and the factors influencing their decisions when WSB is explicitly included as an alternative in their choice set. This gap in the literature motivated the development of this study. This study utilised a structural equation model (SEM) in the light of extended theory of planned behaviour (TPB) to understand the complex interplay of psychological, social and built-environmental factors affecting parents’ intention to use WSB, focusing on the Bradford District, UK. Data on attitude toward WSB, satisfaction on current travel plan, habit, intention to switch to WSB, norms and perceived behavioural control, were collected from 200 respondents. The findings highlight a strong influence of habit and parents’ attitude towards WSB in defining the intention to switch to this new mode. The developed model framework will provide valuable insights for policymakers aiming to promote WSB.

ABSTRACT. Despite ambitious efforts to create better street environments and travel experiences, significant challenges remain. Over three-quarters of 1,072 cities across 120 countries failed to meet the recommended allocation of 20% of urban land for streets and public spaces. Furthermore, many "Healthy Streets" initiatives fall short of addressing a critical question: how do streets influence people's active travel and health? This is particularly pronounced in developing countries, where people-oriented policies and place-based studies are both scarce. Without such evidence, efforts to design effective street environments that promote active lifestyles and improve public health are significantly hindered. This study addresses this gap by examining how street-level features—including safety, comfort, convenience, and sense of community—shape active travel and influence physical and mental health.

A survey conducted in Beijing and a total of 1,747 questionnaires were deemed valid. By adopting structural equation model (SEM), the findings reveal that while Beijing’s street design effectively addresses broad, foundational needs, it often falls short in providing refined, user-specific enhancements, thereby limiting inclusivity. For example, while safety and connectivity for the majority of residents are well-addressed, finer aspects such as barrier-free facilities, direct routes, and noise control remain underdeveloped. This reflects a tendency to prioritize the general needs of most residents while neglecting the specific requirements of vulnerable groups, such as older adults and people with disabilities.

This research also uncovers indirect pathways linking walkability and sense of community to health through active travel, while highlighting age-specific differences in health benefits. Specifically, safety and comfort significantly promote active travel duration, whereas convenience, while shortening the duration of active travel, potentially encourages more physical activities at destinations. A sense of community fosters active travel by strengthening neighborhood interactions, especially among middle-aged and older adults. Young adults experience psychological benefits from active travel, such as stress reduction and emotional regulation, whereas middle-aged and older adults derive greater physical health benefits, including improved physical functioning, chronic disease prevention, and enhanced resilience.

Walking primarily contributes to emotional balance and psychological comfort, helping individuals stay composed and manage smaller, everyday stresses in the moment. In contrast, cycling is particularly effective in aiding recovery, as it builds the capacity to cope with and adapt to more significant hardships and setbacks after challenging experiences. This distinction also implies a differentiated policy focus to address varying types of stress and needs. For walking, policies should aim to enhance the daily walking experience by improving pedestrian infrastructure, increasing green coverage, beautifying streetscapes, and facilitating walking commutes to alleviate small, everyday stresses. Creating walkable communities near residential areas, schools, and workplaces with accessible public spaces such as parks and plazas can further encourage walking as part of daily life. For cycling, policies should focus on expanding mid- to long-distance cycling networks by developing well-connected urban and suburban bike lanes, enabling easier access to longer rides that boost physical activity and psychological resilience. Additionally, promoting recovery-oriented cycling activities, such as organizing endurance rides or outdoor cycling clubs, can help individuals rebuild and adapt after significant stress or challenges.

The analysis further identifies significant disparities in walking durations across different street types: historic district streets encourage prolonged walking due to compact networks and cultural landmarks, urban vitality streets support a balanced distribution of walking durations facilitated by multifunctional infrastructure, and suburban living streets accommodate short trips.

In conclusion, this study strongly supports the integration of street as a key tool for national and local public health policies, establishing optimized street environments as a foundational strategy for building healthier cities.

ABSTRACT. Maritime container terminals are critical nodes in the global supply chain, where stowage planning has significant implications for energy efficiency, operational throughput, and environmental impact. Terminal Planners (TPs) are tasked with translating aggregate stowage plans from the Shipping Line Coordinator (SC) into detailed container assignments, optimizing terminal vehicle operations to minimize travel distances and energy consumption. Efficient stowage planning is a cornerstone of sustainable maritime logistics, as it directly impacts energy consumption, terminal efficiency, and compliance with regulatory standards. This study presents an energy-aware optimization framework for Terminal Planners (TPs), focusing on minimizing vehicle travel distances during container handling. A Mixed Integer Linear Programming (MILP) model is proposed to optimize container placement while integrating key constraints such as ship stability, compatibility, and adherence to the International Maritime Dangerous Goods (IMDG) Code. The process involves two stages: first, a shipping line coordinator creates a general stowage plan, and second, a terminal planner uses a complex optimization model to determine the precise location of each container. The model aims to minimize the distance containers that must travel from the yard to the ship, improving efficiency and reducing energy consumption. This is achieved through an assignment problem, considering various constraints such as container weight, size, type, and safety regulations. The model uses integer and binary variables to represent container placement and incorporates soft constraints to allow for solutions even if all containers cannot be perfectly placed.

ABSTRACT. Despite the potential benefits highlighted in the literature, the widespread adoption of Demand Responsive Transport services remains limited. Integrating passenger and freight transport within the same DRT system could improve its profitability and market suitability, particularly in low-demand areas. The study proposes a model where low-capacity vehicles transport both passengers and parcels within a specific service area, prioritising passenger requests while accommodating deliveries during idle times. This approach aims to optimise resource utilisation and increase revenue without requiring passengers to act as crowdshippers, thereby avoiding administrative and behavioural barriers. An agent-based model is developed to simulate the operation of the integrated service, with passengers and parcels sharing the same vehicles. The model allows for the adjustment of various parameters within a synthetic simulation environment, enabling the evaluation the benefits of prioritising passenger requests compared to scenarios where passengers and freight are given equal priority. By varying input parameters such as the extension of the service area, the ratio between demand densities of travellers and parcels, and the priority discipline adopted during the reservation processes, the results provide useful suggestions for planning integrated passengers-freight DRT services across diverse scenarios, offering practical recommendations for improving the efficiency and viability of such systems.

ABSTRACT. This study explores the potential of crowdshipping as a flexible and technology-driven solution for last-mile delivery by utilizing occasional drivers. Building on previous models like the Vehicle Routing Problem with Occasional Drivers (VRPOD) and the Team Orienteering Problem with Occasional Drivers (TOP-OD), we propose an optimization framework that improves the efficiency of crowdshipping by assigning delivery routes to occasional drivers while balancing reward maximization and travel distance minimization. The results highlight the trade-off between cost and reward, showing significant cost reductions as rewards decrease, with larger instances enabling greater savings. This paper examines how customer distribution patterns influence this trade-off and compares crowdshipping to truck-based delivery, considering the impact of unserved demand on conventional delivery systems.

ABSTRACT. E-commerce deliveries in urban areas have been steadily growing and are facing urban transport planners with new challenges related to flow prediction and scenario assessment (Alverhed et al., 2024; Gao and Zhu, 2022). Most of the current literature on transport flow prediction focuses on e-purchase deliveries from the perspective of transport operators, neglecting the effect of end-consumer choices on operations and city sustainability (Anand et al., 2015; Comi and Delle Site, 2023). Most city-level transport demand models incorporating e-commerce purchases include only a limited number of delivery options for end consumers. Our research examines how end consumers decide on receiving e-purchases, emphasising the impact of the growing range of delivery channels (alternatives) available to them in the new omnichannel context (Comi and Nuzzolo, 2016). Typical new alternatives include crowdshipping, delivery to car trunk, autonomous delivery robots, parcel lockers and staffed third-party pick-up points. These delivery channels are transforming the landscape of last mile delivery and present new opportunities to improve urban transport systems. Understanding how consumers make decisions is a prerequisite for developing predictive urban freight transport models. This paper investigates choice processes by comparing various delivery channels based on service attributes (e.g., delivery time and price, the place where the item is received, etc.), user characteristics and preferences. A stated preference (SP) survey was developed to model delivery mode preferences of consumers. Several studies have examined end-consumer choices concerning specific delivery channels analysed individually: crowdshipping (Cebeci et al., 2023), staffed pickup points (Buldeo Rai et al., 2019) or parcel lockers (Tsai and Tiwasing, 2021). Studies on choice between multiple alternatives are limited. For example, Nguyen et al. (2019) investigated alternatives provided by online retailers such as click and collect service and traditional mail with a main focus on delivery service attributes. Meanwhile, Merkert et al. (2022) compared traditional mail, autonomous delivery robot (drone) and parcel locker in Australia. However, this study characterised alternatives using four attributes and did not consider other innovative delivery channels, such as crowdshipping or delivery to trunk. Furthermore, Wang et al. (2024) examined traditional mail, parcel locker, staffed pick-up point and click and collect in Singapore. The analysis of end-consumer decisions regarding purchase reception was developed from the perspective of willingness to move for collecting items, the effort required for social interactions with staff, and product value. However, the data were collected during the COVID-19 pandemic and end-consumer preferences might have been influenced by the external factors. Based on the literature, there is a quite clear gap in the study of innovative delivery channels compared to well-established ones, as well as a lack of research investigating users’ choice of delivery channels among all available alternatives with various attributes. Therefore, there is need of further research in such field. An SP survey have been developed to investigate end consumers’ choice in receiving e-purchases. The survey, consisting of 4 blocks with 9 hypothetical scenarios, was launched in the second half of last year, yielding 359 responses, mainly from Italy. The first finding of the study confirms that users do not have full knowledge of the alternatives different from the traditional mail. In fact, the analysis reveals two main clusters: one associated with traditional mail and another encompassing all the other delivery channels. A deeper examination of the latter has been performed to investigate how the socioeconomic characteristics influence choices. As a result, a further subgroup has been distinguished, consisting of young and highly educated individuals who are generally more familiar with innovations. Subsequently, this analysis enabled the classification of delivery channels pointing out the set of the service attributes (e.g., destination location, delivery time and price, etc.). Three distinct clusters have been established: (i) traditional mail, (ii) delivery channels already in use such as staffed pick-up points and parcel lockers and (iii) hypothetical delivery ones still in pilot or prototype stages such as autonomous delivery robots, crowdshipping and delivery to car trunk. By examining these classifications, the study highlights the importance of specific characteristics, such as the level of innovation and the requirement for self-collection by the end consumer. Furthermore, the findings provide a valuable tool for forecasting end-consumer decisions on how to receive e-purchases, which should be taken into due consideration when designing new city logistics scenarios. A detailed discussion of the defined clusters and their implications for policymakers is provided in the full paper.

ABSTRACT. This paper proposes a model and procedure for optimising the location of hydrogen refuelling stations for freight vehicles, considering their influence on route choice. Usually, the optimal location of this type of facility is obtained by assuming known and invariable routes followed by the vehicles and by maximising the demand served, i.e. the ‘captured’ flows. The proposed approach assumes that vehicles will adapt their route, which will not necessarily be the minimum-cost route, to reach one of the refuelling stations. The proposed approach will be tried out on a test network to assess its effectiveness.

ABSTRACT. Sustainable travel modes such as electric vehicles (EVs) are considered as one of the main directions for moving towards more sustainable living in urban areas. In this context, this work analyses the influence of main attraction points (such as restaurants, supermarkets, tourist places, schools, hospitals, etc.) on EV charging demand prediction. In designing an EV charging infrastructure plan, factors such as distance or time coverage of EV charging stations can be significant when considering the limited range of EVs. Therefore, starting from a real EV charging station usage dataset of the Lombardy region (Italy), a statistical analysis based on stepwise linear regression has been carried out to estimate how driving distances between charging stations and nearest points of interest (POIs) might affect charging demand. Findings highlight the presence of significant factors in charging demand prediction, which could be useful for further expansions and optimal locations of new charging stations in the Lombardy region.

ABSTRACT. In the pursuit of sustainable urban mobility, electric buses (EB) have emerged as a promising solution to reduce emissions and reliance on fossil fuels. However, the efficient management of their charging schedules remains a critical challenge, particularly for overnight charging (ONC) when buses end their daily operation. In the era of bus electrification, it is imperative to fully replace traditional fossil-fuel buses with electric buses (EBs). This transition requires an increase in bus fleets to sustain service levels due to the limited driving range of EBs. A critical challenge arises in efficiently allocating limited charging facilities to recharge a substantial number of EB batteries overnight in preparation for the next-day operation. This paper provides a mixed integer linear programming (MILP) to formulate the ONC schedule problem. Subsequently, a novel graph-based approach is presented while showing its equivalence to the MILP. By representing charging requests and associated attributes as nodes and arcs, respectively, we leverage graph theory to identify optimal charging sequences that minimize the total required time to fulfill the operational readiness by the next-day morning. In particular, the approach highlights demand-supply considerations, where charging requests are variant with final service times, and the availability of charging facilities is contingent on power grid management. Network flow techniques are employed to automatically derive optimal schedules across various scenarios. The results from multiple instances demonstrate the effectiveness and adaptability of the graph-based approach in comparison to the MILP model. The proposed graph generation procedures offer a scalable solution for transit authorities seeking to exactly allocate charging tasks for the ONC schedule.

ABSTRACT. This study addresses the issue of electric vehicle (EV) overstaying behavior at fast public charging stations by employing a system dynamics and evolutionary game theory approach. Public charging infrastructure is treated as a common-pool resource (CPR), where overstaying EVs can lead to inefficiencies and increased social costs. The research develops a mathematical model to simulate the dynamics of cooperative and non-cooperative EV user behaviors, incorporating factors such as occupation duration, service rates, and social costs. The study compares different charging cost structures, penalties, and social rules to determine their effectiveness in promoting cooperative behavior. Key findings include the superior effectiveness of time-based charging costs over kWh-based costs in reducing social costs, and the efficacy of a charging time limit rule in minimizing inefficiencies. Additionally, the study identifies an optimal State of Charge (SoC) that minimizes social costs in time-based charging cost. These insights contribute to the development of strategies for optimizing public charging station efficiency and reducing the collective burden on EV users.

ABSTRACT. Introduction Driver risk perception (RP) plays a crucial role in influencing driver behavior, particularly in scenarios involving potential collisions or hazardous conditions. RP represents the subjective evaluation of risk based on situational and environmental factors, and it significantly affects the driver’s decision-making process. While past research has highlighted the importance of RP in improving road safety, there is still limited understanding of its dynamic and individualized nature, particularly across diverse demographics and traffic conditions. This study focuses on developing a predictive model for RP using deep learning techniques. Leveraging data from driving simulations, the research investigates the interactions between vehicular trajectories, physiological parameters, and driver demographics. The ultimate goal of the study was to create an adaptive framework that informs Advanced Driver Assistance Systems (ADAS) and enhances overall road safety by accurately modeling and predicting RP in real time. Methodology To analyze RP, the University of Kansas National Advanced Driving Simulator (NADS) miniSim was used. The simulator provided a 170-degree field of view and was equipped with high-definition cameras and an eye-tracking system. The latter captured key physiological metrics such as pupil diameter and eyelid opening, which capture mental workload (WL) and attention levels. The participants included 30 drivers, distributed by gender and stratified into three age groups: 18–24, 25–49, and 50–65 years. This demographic diversity allowed for a comprehensive analysis of RP variability across age and gender. Participants navigated general car-following situations including nine unique risky driving scenarios designed to simulate varying levels of traffic density, aggressive merging, lane changes, and interactions with heavy vehicles. Data collected during the simulations included vehicular trajectories, braking and gas pedal forces, WL fluctuations, and driver characteristics. RP was quantified using a continuous measure derived from braking and accelerator intensity, which served as the response variable for model development. A Deep Learning (DL) algorithm, Long Short-Term Memory (LSTM) neural network was employed as the core predictive model due to its ability to capture temporal dependencies in sequential data. The dataset was divided into training (65%), validation (15%), and testing (20%) subsets to ensure robust model evaluation. The LSTM architecture included 256 hidden units, ReLU activation functions, and an Adam optimizer, with a look-back window of six timesteps to capture short-term RP dynamics. The model performance was assessed using metrics such as Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE). Analysis Results The key research findings include: • Model accuracy: The RP model achieved an RMSE of 0.2 and MAPE of 3.28% which demonstrates strong accuracy in capturing RP dynamics across diverse driving scenarios and demographics. • Role of braking behavior: Braking and gas pedal intensity emerged as a key predictor of RP, with higher forces indicating elevated risk perception. This aligns with prior studies that have identified braking as a primary behavioral response to perceived risk. • Demographic variations: Significant differences in RP were observed across age and gender groups. Younger drivers exhibited more consistent RP patterns but were less sensitive to subtle changes in traffic conditions. Conversely, older drivers displayed greater variability in RP, often influenced by increased WL. Female drivers generally exhibited higher RP levels, particularly in high-density traffic, reflecting heightened caution. • Traffic conditions and RP: RP increased under shorter space headways, lower lead vehicle velocities, and congested traffic. These findings emphasize the importance of situational awareness in shaping risk perception and underscore the limitations of current ADAS in accounting for dynamic traffic environments. • Impact of WL dynamics: Fluctuations in WL significantly influenced RP. Sudden increases in WL, often triggered by unexpected traffic events, enhanced driver alertness and RP. Conversely, sustained low WL levels led to complacency, reducing RP and potentially increasing the likelihood of risky behavior. The LSTM model’s predictions closely mirrored the observed RP values, highlighting its capability to generalize results across diverse scenarios. Additionally, visualizations of predicted versus actual RP values demonstrated the model’s robustness in capturing temporal variations and abrupt changes. Discussion Through the RP model, the driving behavior while approaching various critical traffic conditions including feelings of risk and reaction was explored. The findings have significant implications for enhancing ADAS design and driver training programs. Current ADAS technologies primarily rely on static thresholds and generalizable algorithms, which fail to account for individual differences in RP. By incorporating predictive models such as the LSTM network developed in this study, ADAS can adapt to individual driver profiles and provide tailored feedback or interventions. For example, systems could issue warnings or take corrective actions in scenarios where RP is predicted to be insufficient, potentially preventing collisions. Similarly, integrating real-time RP predictions into ADAS could improve situational awareness by dynamically adjusting system sensitivity based on driver state and environmental conditions. The study also underscores the importance of demographic-specific interventions. Younger drivers, characterized by lower RP sensitivity, could benefit from training programs focusing on hazard identification and risk assessment. Conversely, interventions for older drivers could address variability in RP and enhance their ability to respond to complex traffic scenarios. Gender-specific insights, such as the heightened RP observed in female drivers, could inform the development of more inclusive and effective safety measures. The integration of RP models into traffic safety frameworks has the potential to inform policy and infrastructure design. For instance, urban planning initiatives could leverage RP data to identify high-risk zones and implement targeted interventions, such as optimized signage, reduced speed limits, or enhanced road markings. Additionally, insights from RP models could support the development of advanced simulation-based training programs that replicate real-world driving conditions and improve driver preparedness.

Conclusions and Future Directions This study presents a novel approach to modeling RP using deep learning, offering valuable insights into the cognitive and behavioral aspects of driving. By leveraging driving simulation data, the research highlights the dynamic and individualized nature of RP, providing a foundation for developing adaptive systems that enhance road safety. The LSTM model’s strong performance underscores its potential as a tool for real-time RP prediction, with applications spanning ADAS design, driver training, and policy. Future work will explore the integration of RP predictions into real-world ADAS, emphasizing adaptive feedback and intervention mechanisms. Future research should focus on validating the model in real-world driving conditions, incorporating additional physiological and psychological parameters, and expanding the demographic scope. Integrating additional data sources, such as heart rate variability, Electroencephalogram (EEG) signals, and external environmental factors, such as weather, could further enhance the model’s accuracy and applicability. Furthermore, extending the analysis to include longitudinal data could provide insights into how RP evolves over time and across varying traffic contexts. By bridging the gap between theoretical research and real-world applications, this work contributes to the ongoing effort to create safer, more adaptive transportation systems for all road users.

ABSTRACT. Physics-based car-following models are widely used for microsimulation and predictive control purposes. However, human driving behaviors deviate from machine-like driving described by such models because of intra- and inter-personal heterogeneity. Purely data-driven car-following models have been proposed as a powerful alternative but may show poor performance when sufficient data is not available. Hybrid physics-constrained or physics-informed deep learning models link domain knowledge with machine learning that could handle complex traffic conditions and therefore produce better trajectory estimates. In this study, we propose a novel hybrid model in which a Bayesian convolutional neural network (CNN) is employed as a complementary computation layer to improve the accuracy of physics-based models. As such, the Bayesian method produces a range of possible outputs rather than predicting a single-point estimate that can be used to measure prediction uncertainties (or model's confidence level). We tested our approach on a simulation dataset with stochastic time-dependent artificial noise and a real-world dataset (NGSIM). The experimental results showed that our model could learn not only the systematic errors of physics-based models, but also the heterogeneity in human driving behaviors. As a result, we achieved considerably more accurate velocity estimates and, thereby obtaining accurate vehicle trajectories.

ABSTRACT. Accurate traffic prediction has drawn increasing attention in recent years due to its potential to optimize traffic operation efficiency. Recent studies mostly combine Graph Neural Network (GNN) and Recurrent Neural Network (RNN) for traffic prediction. However, existing methods have not explicitly analyzed how the spatio-temporal dependencies of surrounding roads contribute to the traffic flow of a given road segment, potentially leading to a spatio-temporal lag effect. Specifically, the traffic flow time series of the adjacent upstream and downstream road segments show a high similarity, but a phase difference due to traffic propagation. This characteristic could facilitate traffic prediction for road network. To capture this pattern, a hybrid framework, named Dynamic Time Warping based Spatio-Temporal Graph Neural Network (DTW-ST-GNN), is proposed. Firstly, this framework utilizes Dynamic Time Warping (DTW) to align the time series data and capture the hidden spatial correlations, which are then used to generate a dynamic adjacency matrix. Secondly, the traffic features of the road segments are aggregated by GNN to investigate the spatial dependencies. Finally, time series model is applied as a component of the model to extract the temporal dependencies. The proposed model is evaluated through carefully designed experiments based on real-world data.

ABSTRACT. This work presents a novel solution to the issue of traffic forecasting in sensorless urban areas using Floating Car Data (FCD). FCD is scalable and widespread and offers the potential for accurate traffic flow estimates for areas without fixed traffic sensors through the use of machine learning algorithms. Using FCD and a limited set of fixed sensors for training, a forecasting model is developed, enabling the estimation of traffic flows on major urban roads up to four hours in advance. This method allows traffic managers to forecast conditions throughout the network at a lower cost than using physical sensors. Preliminary results from the case study of Catania (Italy) suggest that the proposed methodology has the potential to improve traffic management with limited investment costs.

ABSTRACT. The increasing demand for road transport in Europe has raised concerns regarding sustainability, as evidenced by its significant environmental and economic impacts. Since 1995, cargo transport within the EU has risen by 40%, with over 73% of freight transported by road. This surge contributed to a 24% increase in greenhouse gas emissions from the transport sector by 2019, accounting for 31% of total EU energy consumption, predominantly sourced from road transport. In response, the European Green Deal aims to promote alternative transport modes, with a particular focus on enhancing inland waterway transport (IWT), notably in underutilized regions like Lithuania. Despite its strategic geographic position and extensive hydrographic network, Lithuania’s IWT remains underdeveloped due to inadequate infrastructure and navigational challenges. This study presents a multi-objective, multi-commodity, multi-modal transportation model aimed at optimizing cargo allocation across various transport modes, particularly transitioning from road to IWT. Key objectives include minimizing operational costs and maximizing sustainability through reduced CO2 emissions. Through a case study involving a local logistics provider, the research will assess the model’s effectiveness in improving transport efficiency, identifying optimal routes and transshipment points while aligning with broader sustainability goals. The outcomes are anticipated to yield actionable insights for enhancing IWT utilization, ultimately fostering a more integrated and sustainable transport network in Europe. By advocating for a balanced approach that incorporates economic viability alongside sustainability, this study contributes vital knowledge for transport planners and policymakers navigating the complexities of modern logistics.

ABSTRACT. Micromobility services have expanded rapidly, causing challenges like congestion and improper parking. In response, cities have introduced regulations to balance the interests of users, providers, and non-users. However, current research neglects an empirical evaluation on the impact of such regulations. This study evaluates the impact of micromobility policies in three German cities—Bremen, Frankfurt, and Nürnberg—using the Difference-in-Differences (DiD) approach. The treatment cities were subjected to fleet size and parking zone policies. The analysis investigates the total number of trips and the average number of trips per vehicle per day, utilizing data obtained from API queries of seven micromobility operators between March 2023 and July 2024. The findings suggest that fleet size limitation leads to a significant reduction in total trips, but does not affect the average trips per vehicle. In contrast, parking zone policies showed no significant effects on either the number of trips or vehicle utilization. Our results inform city municipalities how to regulate micromobility without harming the overall operational effectiveness of micromobility operators. Instead, city municipalities can regulate micromobility to balance the needs of non-users, e.g., reduction of improper use, with those of operators. These insights are crucial for balancing urban planning needs with provider viability.

ABSTRACT. In the effort to mitigate climate change, Indonesia is stimulating the introduction of electric motor-cycles (EMs). The present study aims to contrast preferences for adopting EM in two cities in Indo-nesia, namely Jakarta and Yogyakarta, by means of a stated preference survey with 894 respondents overall. This study finds that purchase price is one of the most important determinants for adopting EM. Comparing these two cities with distinct socio-economic characteristics, the study found that people in Jakarta have higher WTP than people in Yogyakarta in all key attributes, such as driving range, charging time, and charging station. This disparity suggesting government to develop tailored policies which area-based specific to promote EM in different area, such as higher purchase subsidy in lower WTP city and more fast charging stations in higher WTP city

ABSTRACT. Valence framing, where equivalent information is presented in either a positive or negative light, is a nuanced topic that has previously not been well accounted for in choice modelling literature. This may explain inconsistent results within the field and findings contrary to other fields, such as behavioural economics, where framing definitions are typologised. We endeavour to bridge this gap across fields by introducing a valence framing typology into a discrete choice experiment in a transport context, in an attempt to show that different types of framing should be more prominent within choice modelling because they can have differing results. Furthermore, a discrete choice experiment allows for the possibility of these three valence framing types to be investigated simultaneously, in one, multi-attribute decision context, which has not been explored previously. We ask participants to complete a series of choice tasks where attributes are framed using different valences and different types of framing. We gain insights into participants' risk preferences and we find a relationship between response duration and frame condition, although no main framing effects are identified. We also consider the potential reasons for this and explore further avenues for analysis.

ABSTRACT. Cost pass-through plays a central role in understanding how airlines respond to external shocks, from airport charges to air passenger taxes. When a cost increase applies to a number of carriers equally - such as an air passenger tax - each airline faces similar incremental costs, reducing the risk of losing market share by raising fares. Sector-wide cost increases therefore tend to cause higher pass-through. By contrast, a full pass-through of localized or airline-specific cost shocks would generate price differentials relative to non-affected airlines, thereby reducing the degree of pass-through (Koopmans & Lieshout, 2016). For example, if airport charges rise at a specific hub, carriers that primarily operate from that hub may face a greater burden of the cost. This could happen because competing airlines with alternative bases might be able to offer lower prices for the same destination (Brueckner, 2002). Understanding differences in pass-through is crucial for developing effective regulatory policies and for guiding airport charge policies that balance competition, connectivity, and sustainability goals.

The objective of this paper is to empirically test the hypothesis that sector-wide cost shocks exhibit a higher pass-through on airfares than idiosyncratic cost increases. More specifically, the study estimates how airfares change in response to cost shocks of varying scope and intensity and whether this interacts with market concentration. For identification, I make use of variation in airport-specific charges such as landing charges and country-wide shocks such as air passenger taxes across European airports between 2018 and 2024, while excluding 2020–2022 due to the predominant impact of COVID-19. I apply a two-way fixed-effect regression to isolate the effects of sector-wide and firm-specific cost shocks on airfares. For example, air passenger taxes - simultaneously affecting all airlines - serve as a treatment for sector-wide shocks. In parallel, I regress airfares on airport-specific charges such as landing fees to identify idiosyncratic shocks. This reveals the magnitude of pass-through while controlling for time-invariant route characteristics, airline operational differences, and wider demand trends. To capture heterogeneity in pass-through, I include an interaction term which measures market concentration.

This study makes use of the Sabre Market Intelligence (Sabre-MI) data set, which provides comprehensive monthly data on airline bookings. Sabre is the primary provider of the global airline ticket distribution system used by airlines for sales processing. Sabre-MI offers three key features that are crucial for the analysis. First, Sabre-MI includes reliable global transaction data on airfares. Second, the panel structure of the data set allows to control for time-invariant factors. Third, the data set provides monthly observations of origin and destination airport-pairs, separated by airline, thus allowing for heterogeneous analysis under different market concentration. Information about taxes and charges at airport level is provided by RDC on a monthly basis.

The preliminary results show that sector-wide costs lead to higher pass-through compared to airport-specific costs, confirming the prevailing theoretical argument that airlines more freely pass along universal charges. In line with theoretical priors in oligopolistic markets, pass-through decreases with market concentration. The paper contributes to the air transport literature by providing empirical evidence on the differences in pass-through between sector-wide and airline-specific cost changes. By quantifying the magnitude of the effect, the study adds to the understanding of how policymakers can optimize transportation networks and promote sustainable travel options. Greater pass-through of sector-wide taxes may be welcomed by policymakers seeking to internalize environmental externalities. Finding that localized airport fees exhibit lower pass-through suggests that these fees could create competitive disadvantages for certain airports or airlines.

References

Brueckner, J. K. (2002). Airport congestion when carriers have market power. American Economic Review, 92(5), 1357–1375. Koopmans, C., & Lieshout, R. (2016). Airline cost changes: To what extent are they passed through to the passenger? Journal of Air Transport Management, 53, 1–11.

ABSTRACT. The widespread availability of passive GPS data from trucks presents new opportunities for analyzing freight operations and demand patterns. However, despite the growing body of research, there is still a need for more comprehensive studies that explore large-scale freight movements across diverse metropolitan areas. This study addresses this gap by utilizing a nationwide dataset of truck movements in Italy and applying machine learning techniques to identify operational patterns in selected urban contexts. By linking freight activity to urban characteristics, the analysis provides empirical insights to support urban freight planning, infrastructure development, and regulatory policies. The integration of unsupervised learning models with descriptive statistics establishes a robust methodological framework for processing large-scale GPS data, offering a transferable approach for similar studies in other countries.

ABSTRACT. Purpose of this paper: The increasing demand for sustainable food logistics requires developing a cold chain model with zero emissions which uses energy-efficient technologies along with renewable energy sources and digital advancements. This study examines how to develop and implement a sustainable transportation approach for perishable food logistics in Scotland with a special focus on beef supply chains. The project tackles crucial cold chain logistics challenges with advanced refrigeration systems and alternative energy solutions combined with enhanced vehicle technologies to reduce emissions while maintaining food security and operational effectiveness at reduced costs. The proposed framework focuses on transforming cold chain infrastructure to reach energy efficiency targets and employ technological advancements together with economic optimization while ensuring the secure and efficient movement of perishable products. The study aims to: • Revise and update existing energy consumption information and CO₂ emission data related to cold chain logistics systems. • Evaluate the cooling requirements for transporting fresh produce to reduce energy usage. • Analyse future cooling energy requirements and their effects on the resilience of the system. • Propose a beef cold chain model designed to minimize costs and environmental impact while integrating decarbonization techniques and energy-efficient methods along with digitalization strategies to decrease post-harvest losses and emissions. This research provides strategic guidance to roadmap for industry stakeholders, policymakers, and researchers to implement sustainable, low-carbon solutions in cold chain management. The study highlights innovative strategies, technologies, and models to ensure food security, affordability, and environmental sustainability within the cold chain.

Design/methodology/approach: The study applies an optimization framework based on simulation techniques and Agent-Based Modeling (ABM) which follows the CRIO (Capacity, Role, Interaction, and Organization) metamodel structure. The CRIO framework serves as a structured approach in ABM and simulation design specifically for complex system modeling centered around groups and roles as fundamental components. Through its systematic methodology it enables organized establishment of capacities and roles while defining interactions and organizational structures to boost efficiency and meet net-zero objectives in cold chain management. The methodology follows these key steps: 1. System Representation: Create a foundational model of the existing cold chain system that depicts the current operations by including essential agents such as farmers and suppliers along with other stakeholders like distributors and retailers and logistics elements including ports and transportation vehicles then analyze their operational behaviors and emissions. 2. Strategy and Technology Identification: Identify a variety of strategies and technological advancements to achieve zero emissions in the cold chain through renewable energy adoption, as well as other approaches such as vehicle electrification, transportation route optimization, low-carbon refrigeration technology utilization, and digital transformation. 3. Data Collection: Collect real-world UK beef supply chain data then scale it down to represent Scotland while focusing on energy consumption levels and emission production along with geographical distribution of supply chain agents, transportation distances and logistics parameters. 4. Scenario Analysis: Simulate a set of feasible scenarios in Scotland that combine different energy sources such as biogas and renewable energy with vehicle technologies like electric vehicles and different refrigeration systems and evaluate how transitioning to a digital auction market affects supply chain emissions and efficiency. 5. Simulation Outputs and Visualization: Generate customizable charts that allow users to analyze energy consumption patterns against emissions and estimated costs while comparing different scenarios. Our organized method facilitates decisions based on data analysis by providing an extensive review of sustainable approaches needed to achieve a net-zero cold chain system.

Findings: The study shows how renewable energy sources along with electric vehicles and advanced refrigeration technologies together with digitalization can lead to substantial emission reductions. The study revealed better decision-making abilities alongside decreased food waste and more effective cold chain operations. The simulation demonstrates how feasible it is to move toward sustainable logistics models and reveals detailed pathways toward zero emissions. Furthermore, the proposed model demonstrates its scalability over various years and market conditions which enables broader implementation beyond the beef cold chain. The model demonstrates adaptability across different perishable food supply chains while confirming its usefulness as a decision-support tool for sustainable logistics planning. This research offers essential knowledge that helps policymakers and industry executives alongside supply chain stakeholders to establish a sustainable cold chain system that balances low carbon emissions with energy efficiency and economic viability.

Value and Practical implications: This paper provides a comprehensive decision-support framework for stakeholders in the cold chain industry and connects academic research with real-world application. By integrating advanced modeling techniques, it offers valuable insights to facilitate data-driven decision-making in sustainable logistics. Key contributions of this research include: • Policy Development: The research develops policy tools that enable policymakers to design incentive structures that promote renewable energy adoption and support sustainable supply chain transitions. • Industry Transformation: We deliver strategic guidance to logistics providers about shifting towards zero-emission fleets, reducing carbon footprints, and improving operational efficiency. • Academic and Industry Insights: The research enables both academics and industry practitioners to use simulation-based methods for assessing sustainability strategies within cold supply chains. • Investment Justification: The paper shows how long-term savings justify investments in electric vehicles, modern refrigeration systems, and digital innovations for private sector funding. This research aids the UK’s 2050 net-zero goals while establishing a standard for worldwide sustainable logistics operations. This research provides innovative strategies that achieve economic sustainability and environmental objectives while promoting partnerships among government agencies and leaders of industry and technology to advance sustainable changes in the cold chain sector. The results provide stakeholders with actionable recommendations to improve resource distribution and deploy cutting-edge logistics approaches while introducing low-emission technologies. This research provides policymakers with a strategic framework which will guide the development of regulatory measures that enable the UK cold chain sector to reach its net-zero emissions goal by 2050 and keeping the economy stable and making sure that operations run smoothly.

ABSTRACT. This paper introduces a new probability-based model for simulating goods delivery demand in urban areas with traffic restrictions. Building on the four-step travel demand model, our approach generates demand as a flow of cargo delivery requests, differentiated by client type. We apply the model to Krakow’s Old Town, a zone where traffic of conventional vehicles is restricted, and implement it in Python. Our analysis, applied for the case of simulating deliveries by cargo bikes inside the studied urban area, demonstrates that while request probabilities influence individual client demand, they do not significantly affect the aggregate simulated demand. This methodology offers a practical tool for transportation planners seeking to optimize logistics in congested urban centers.

ABSTRACT. 1. Introduction

In an increasingly globalized world, the economic performance of countries is becoming ever more dependent on foreign trade. This is especially true for export-oriented economies like Germany. Sea and inland ports provide play a key role by providing industries with access to international good markets for export of products and import of intermediate inputs. In doing so, ports support employment and value added in the economy.

However, ports contribute not only to national employment and GDP by acting as enablers for other industries but also through their own economic activities, known as direct effects. In addition, ports purchase intermediate inputs such as energy, which stimulates economic activity in supplying industries, resulting in indirect effects.

Several studies have explored the employment and value-added effects of ports in Germany, with research ranging from analyses of individual ports to broader evaluations of the maritime economy (HPC, 2018; ISL, 2017; ISL et al., 2019; ISL, ETR, Fraunhofer CML, and DIW Econ 2021; ISL, ETR, Fraunhofer CML, and Ramboll, 2021). However, all of these studies rely on data from 2018 or earlier and therefore do not reflect the lasting impact of the COVID-19 pandemic. Furthermore, the potential consequences of the political commitment to net-zero emission targets for employment and value added are largely unaddressed.

This paper estimates the employment and value added directly and indirectly generated by German ports in 2022. In addition, it analyzes the factors influencing port employment and value added up to 2040, including the expected sharp decline in fossil fuel imports.

2. Methodology and Data

The economic effects of ports include direct and indirect effects. Direct effects refer to the employment and value added generated through the activities of the port stakeholders themselves. Data on economic activities at ports are largely obtained from official statistics published by the German Federal Statistical Office as well as industry reports. The industry sector “Cargo handling” in the official statistics is not differentiated by transport mode, so we estimate the share attributable to ports based on the proportional handling volume.

Indirect effects refer to the employment and value added generated by the economic activity of the suppliers of the port stakeholders. These effects are thus the result of the purchase of intermediate inputs by firms located at the port. To estimate the indirect effects, we first determine the amount of intermediate inputs purchased by the port stakeholders.

To model the employment and value-added effects in the supply chain based on the purchase of intermediate inputs, we use an input-output model and data from an input-output table. An input-output table represents the flow of goods and services between different industries within an economy. For Germany, we use the national input-output table from the German Federal Statistical Office for 2019.

The result of the input-output model is a set of multipliers, which indicate how much employment and value added are generated by a €1 million purchase of intermediate inputs. The indirect effects are then calculated by multiplying these multipliers with the determined intermediate input purchases.

The potential future development of the economic impacts of ports is analyzed based on the “Transport Forecast 2040” for Germany (Intraplan et al, 2024), which includes projections on cargo throughput and cargo types at sea and inland ports.

3. Results and Conclusions

The analysis reveals that sea and inland port contribute 60,500 jobs and €6.1 billion euro to the German economy in 2022. This includes both direct economic activity at ports and indirect economic activity in the upstream supply chain. The input-output model results indicate that for every €1 million spent by port stakeholders on intermediate inputs in a year, 5.3 jobs are created in the supply chain, along with €387,000 in value added. Furthermore, the assessment based on the “Transport Forecast 2040” suggests that while total cargo volumes at German sea ports are projected to grow marginally, total cargo volumes at German inland ports are anticipated to decline by about 16%. This poses a significant challenge for inland ports and could lead to a reduction in employment and value added at the affected ports.

Given these finding, policy makers and port stakeholders are required to take proactive measures to ensure long-term employment and value added at ports, in particular at inland ports facing declining cargo volumes. For inland ports, this could involve diversifying port activities such as strengthening integration with rail and road transport. Additionally, both inland and sea ports are required to continuously enhance operational efficiency, for example by introducing or expanding automated cargo handling systems.

References

HPC, 2018. Ökonomische Effekte des Kieler Hafens - Ergebnisbericht [Economic effects of the port of Kiel – Final report]. Study commissioned for the Seehafen Kiel GmbH & Co. KG.

Intraplan, Trimode, ETR, MWP, Fraunhofer CML, 2024. Verkehrsprognose 2040 – Band 6.1 E: Verkehrsentwicklungsprognose: Prognosefall 1 „Basisprognose 2040“ (Ergebnisse) [Transport forecast 2040 – Volume 6.1 E: Transport development forecast: Forecast case 1 „Base forecast 2040“ (Results)]. Commissioned by the Federal Ministry of Digital and Transport (BMDV). VB970423.

ISL, 2017. Beschäftigungseffekte der bremischen Häfen für das Jahr 2015 – Zusammenfassung [Employment effects of the ports of Bremen for the year 2015 – Summary]. Study commissioned for the bremenports GmbH & Co. KG.

ISL, ETR, Fraunhofer CML, DIW Econ, 2021. Maritime Wertschöpfung und Beschäftigung in Deutschland [Maritime value added and employment in Germany]. Study commissioned by the Federal Ministry for Economic Affairs and Energy (BMWi).

ISL, ETR, Fraunhofer CML, Ramboll, 2021. Volkswirtschaftliche Bedeutung des Hamburger Hafens – Untersuchung der regional- und gesamtwirtschaftlichen Bedeutung des Hamburger Hafens – Endbericht [Economic significance of the port of Hamburg – Analysis of the regional and macroeconomic significance of the port of Hamburg – Final report]. Commissioned by the Hamburg Port Authority.

ISL, Fraunhofer CML & IML, ETR, Holocher, 2019. Untersuchung der volkswirtschaftlichen Bedeutung der deutschen See- und Binnenhäfen auf Grundlage ihrer Beschäftigungswirkung [Study on the economic significance of German sea and inland ports based on their employment effects]. Commissioned by the Federal Ministry of Transport and Digital Infrastructure (BMVI).

ABSTRACT. Coffee is one of the most consumed beverages worldwide, driving an industry worth billions of euros annually. However, its life cycle does not end with consumption; it generates a significant amount of by-products, including spent coffee grounds (SCG), often treated as waste. If managed through innovative recovery and reuse practices, these by-products can become valuable resources, reducing environmental impact and optimizing logistics processes. This research explores the recovery of SCG as secondary raw materials in other production chains, leveraging IoT technologies and web-based platforms for real-time monitoring and informed decision-making. The study proposes an algorithm that simultaneously plans coffee delivery routes and SCG recovery using an integrated approach, while dynamically adapting to evolving requests through a Rolling Horizon technique. Performance results on test cases based on extensive territorial and realistic consumption data are promising, highlighting the potential of this approach for sustainable supply chain management.

ABSTRACT. Urban transport systems play a crucial role in addressing sustainability challenges. While public awareness of environmental issues has increased, there remains a persistent gap between climate-conscious attitudes and actual travel behavior. This study investigates the role of environmental attitudes and zero-emission technologies, specifically zero-emission buses (ZEBs), in shaping sustainable mobility intentions in the United Kingdom. Using data from the UK National Travel Attitudes Survey (NTAS), this study examines the perceived importance of environmental factors in shaping travel behaviour. In addition, the study assesses the factors influencing individuals' willingness to shift toward public transport, particularly in response to the introduction of ZEBs. Ordered probit models are estimated considering socio-demographic characteristics, environmental awareness, financial resources, accessibility, social norms, and behavioural characteristics. The results indicate that while nearly two-thirds of respondents recognise environmental considerations as important in their transport choices, only 35% express an increased likelihood of using public transport if ZEBs were implemented. Frequent public transport users demonstrate higher environmental awareness and willingness to shift, whereas car-dependent individuals remain resistant. Socio-demographic factors also play a role, with females and elderly individuals more inclined toward sustainable travel, while high-income individuals exhibit lower interest in public transport despite ZEB adoption. The findings highlight the potential of ZEBs to encourage public transport use but also underscore the necessity for targeted policies that address accessibility, cost, and service reliability.

ABSTRACT. This study examines city-wise road traffic noise emissions at link and district levels across through automated traffic planning. Using Rapidex traffic data, noise levels are modeled based on traffic volume and vehicle composition. To assess disparities in noise exposure, mean excess noise and Gini equality index is computed for evaluating district-level maximum and mean noise exposure and disparities. Results reveal disparities across districts highlighting the need for equitable noise mitigation strategies targeting socio-economic vulnerabilities.

ABSTRACT. This paper analyzes control strategies to prevent and mitigate the effects of loss of headway adherence in a bus line. A microsimulation model is created to replicate the performance of an electric bus line in Barcelona, and three different strategies are applied: speed reduction, traffic light priority and stop skipping. Strategies are applied individually and combined, with different activation thresholds. The indicators used to assess the strategies include the commercial speed, the coefficient of variation of the headway, and the waiting and in-vehicle travel time of the passengers. Results indicate that a combined strategy with both speed reduction and traffic light priority improves the coefficient of variation of the headway by 27%, while reducing the commercial speed 2.2%. The model created allows to evaluate multiple cases, such as events, road works, or to include new strategies.

ABSTRACT. 1. Introduction Unobservable feelings are essential drivers in individuals' travel choices. Research has linked subjective experience (Lim et al., 2024), satisfaction (De Vos et al., 2022), and mood or emotion (Böcker et al., 2016) to how people choose travel modes. These influences are typically examined through stated self-reports, which can be prone to recall error and report bias. They can also be distorted by memory bias and other cognitive factors (remembering only the most intense or recent experiences) , leading to potential mismatches between what travellers actually experience and what they report. Consequently, such methods may not fully capture the emotional states that shape travel behaviour. In recent years, various fields have begun using physiological sensor data to measure these unobservable states (Whyte et al., 2024; Richardson et al., 2020). Signals such as heart rate, electrodermal activity, and skin temperature can reveal emotional changes without relying solely on self-report. Transportation researchers have also adopted these methods to see how stress or mood affects travel choices (Paschalidis et al., 2019; Chen et al., 2017). Most of these are however laboratory studies, which offer precise control of external variables, but may not reflect real-world complexities (Hancock & Choudhury, 2023). Traffic congestion, delays, or social interactions can alter emotional states in ways that differ from controlled settings. Collecting physiological data in the real world poses methodological challenges. Fully naturalistic data can carry excessive “noise,” making it hard to link specific emotional shifts to particular trip features. In addition, the choice and consideration sets are often unobserved. Although Henriquez-Jara (2025) carried out a field experiment to collect physiological data, participants in that study did not engage in choice behaviour. Moreover, unobservable factors like non-travel stress (e.g. work deadlines, arguments, family crisis) can overshadow the feelings linked to a journey. These issues highlight the need for compromise between laboratory control and real-world relevance if we are to better establish the role of these factors in travel behaviour choices. In this study, we designed a controlled field experiment to investigate how mood and experience affect travel mode choice. We limit the choice set to a few modes and destinations, keeping journeys within controlled conditions. We plan to collect both physiological sensor data and self-reported experiences, addressing three research questions:(1) Does physiological data match self-reported travel experiences in real world settings? (2) How do prior experiences shape current ones via anchoring or reference points? (3) Do travellers' unobservable feelings directly affect subsequent mode choices, or do they change how time, cost, and comfort are weighed in the utility of the modes? By answering these questions, our study will shed new light on how mood and prior experiences influence travel decisions beyond what self-reports alone can reveal. 2. Experiment design We have designed an incentive-compatible travel experiment in Leeds. Participants will complete a multi-stage journey through four destinations, labelled A→B→C→D→A (as shown in Figure 1). They must finish the entire activity within a set time limit (2 hours). At the start, each participant will receive a travel budget (£30) that they can spend on fares. They are free to choose among bus, train, and taxi at each stage. Any unspent budget remains theirs as a reward, so the choices have real consequences and there is a clear incentive to balance time and cost. As seen in Figure 1, before the journey begins, participants stay in a quiet environment at location A to record baseline physiological signals. During this baseline, they also complete a stated-preference (SP) experiment to measure their initial travel preferences. Once the trip starts, they wear an Empatica Embrace wristband, which measures indicators such as electrodermal activity, heart rate, and skin temperature in real time. This device allows us to capture momentary changes in stress or emotional state without relying solely on self-reported data. Upon reaching each destination (B, C, and D), participants fill in a short questionnaire about their experiences during the previous stage (e.g. A→B, B→C, C→D). They report subjective mood, perceived comfort, and overall satisfaction. They then use Google map to check the travel details of the next leg. For example, they can see typical cost, travel time for the bus, train, and taxi options. They record these attributes, as well as any immediate impressions, and make their next mode choice. This process continues until they return to A. At this point, the participants will be asked to complete another SP experiment to test for preferential change. By combining the sensor data with repeated subjective measures and actual decisions, we can build a rich dataset that includes physiological responses, reported experiences, travel attributes, and chosen modes at every stage, and stated preference choice data. The target sample size is 150 individuals (Observations SP1: 5 each, RP: 4 each, SP2: 5 each, 2100 observations in total). 3. Proposed model We propose to develop a dynamic choice model in which each leg's perceptions and experiences feed into the next stage's decision process (Wang et al., 2025). As shown in Figure 2, travel attributes (e.g. cost, time) and environment factors (e.g. weather, time pressure) combine to form a utility function for each stage. The utility in one stage is influenced by physiological and stated experiences from the previous stage, capturing how mood or stress might carry over. For example, a stressful taxi ride from A to B could lower a participant's willingness to choose taxi again from B to C. Each choice then leads to a new experience, which becomes input for subsequent decisions. Over the four-stage journey, this sequential structure reveals how experiences and feelings can shape evolving preferences and final choices. Through this model, we aim to identify both immediate and lasting effects of physiological and subjective states on travel decisions. 4. Expected Findings and Contributions The main contribution of this research will be the integration of real-world physiological data with dynamic choice modelling, revealing how momentary emotional states accumulate throughout a trip. We expect to uncover how physiological states and self-reported feelings influence travel decisions across multiple stages. Specifically, we aim to explore the lasting effects of unobservable feelings, as well as possible mood “anchoring” on subsequent choices. Overall, the study is expected to bridges the gap between laboratory studies and self-reported data and advance our understanding of the influence of mood and experience on travel behaviour.

ABSTRACT. This paper examines the impact of trip characteristics and socio-demographic factors on bike trip duration in three European cities: Tallinn, Braga, and Istanbul, using a Latent Class Accelerated Failure Time (LCAFT) model. The model reveals that socio-demographic factors, such as gender, age, and education, influence travel patterns differently across these cities. In the data from Tallinn and Braga, both age and gender significantly affect trip duration. In contrast, in Istanbul, age and education level are the more decisive factors. These findings underscore the importance of considering the specific data contextual factors across different geographic locations. Our elasticity analysis further demonstrates that the impact of various factors on bike trip duration varies significantly by city.

ABSTRACT. Introduction

Uncontrolled intersections comprise significant safety challenges, particularly in countries like Sri Lanka, where mixed traffic conditions and the absence of explicit priority regulations create a high-risk environment (Vajeeran & De Silva, 2020). At these intersections, drivers must rely on their own judgment to select appropriate gaps in traffic when making turning or crossing maneuvers. Improper gap acceptance decisions often lead to vehicle conflicts, increasing the likelihood of accidents (Bhatt & Shah, 2022). Among vulnerable road users, motorcyclists and three-wheeler drivers face higher risks due to their maneuverability and relatively smaller vehicle sizes. One of the main challenges at unsignalized intersections is the lack of a structured right-of-way system, which forces drivers to make quick decisions under varying traffic conditions. Unlike in fully signalized intersections, where traffic lights regulate movement, unsignalized intersections depend entirely on driver behavior and judgment (Sandaruwan, Karunarathne & Wickramasinghe, 2019). This results in frequent miscalculations, particularly in high-traffic environments, where drivers must negotiate right-of-way with oncoming vehicles. In such situations, gap acceptance plays a crucial role in determining traffic flow and safety.

Several factors influence gap acceptance behavior, including traffic density, vehicle speed, road geometry, driver experience, and environmental conditions (Ashalatha & Chandra, 2011). Understanding these factors is essential for developing effective road safety measures. In mixed-traffic conditions, motorcyclists and three-wheeler drivers often adopt different strategies when selecting gaps compared to drivers of larger vehicles. These differences arise due to variations in vehicle acceleration capabilities, visibility, and perceived risk. Identifying these behavioral patterns can help traffic engineers design safer intersections by incorporating strategies that accommodate the needs of all road users. This research aims to analyze the factors influencing gap acceptance behavior at unsignalized T-junctions, determine the critical gap sizes typically accepted by local drivers, and examine variations in decision-making patterns based on vehicle type and driver characteristics. By identifying these factors, the study contributes to the development of safety strategies and intersection design improvements tailored to local traffic conditions. The findings provide insights for urban planners, policymakers, and traffic engineers in designing safer and more efficient road infrastructure.

Methodology

To gain a detailed understanding of gap acceptance behavior, field observations were conducted at 20 unsignalized T-junctions across different urban locations in Sri Lanka. These intersections were selected based on traffic density, road geometry, and the presence of diverse vehicle types, ensuring a representative sample of real-world conditions. Data collection was carried out during peak and non-peak hours to capture variations in driver behavior under different traffic volumes. A video recording method was used to monitor traffic movements at each intersection. Cameras were positioned at strategic locations to capture clear footage of vehicle interactions, ensuring visibility of gap selection decisions made by drivers. The collected footage was later analyzed to extract key parameters, including the arrival rates of vehicles, the gaps accepted or rejected by drivers, and crossing times at the intersections. Additionally, driver characteristics, such as gender and approximate age, were noted where possible to examine their influence on gap acceptance behavior.

In addition to traffic behavior, road conditions and environmental factors were documented. This included factors such as visibility at intersections, pavement conditions, weather conditions, and road markings. These variables were considered to determine whether external conditions played a role in influencing a driver's decision to accept or reject a gap. To estimate the critical gap, Raff’s method was applied. This statistical technique identifies the minimum time interval a driver perceives as safe for crossing or merging into traffic. The method is widely used in traffic engineering to analyze gap acceptance and helps in establishing baseline values for intersection safety improvements. The extracted data was then processed using Python’s Pandas and NumPy libraries, enabling efficient analysis of large datasets. Statistical modeling was used to identify patterns and variations in driver behavior, ensuring a comprehensive understanding of how different factors influence gap acceptance decisions.

Results

The findings revealed significant differences in gap acceptance behavior among various categories of road users, highlighting how different vehicle types and driver characteristics influence decision-making at unsignalized intersections. Motorcyclists generally exhibited a more aggressive driving style, accepting smaller gaps compared to three-wheeler drivers. On average, motorcycles accepted gaps as short as 1.76 seconds, whereas three-wheelers waited for slightly longer gaps, around 1.81 seconds, before crossing or merging into traffic. This suggests that motorcycles, due to their greater maneuverability and smaller size, tend to take more risks when entering intersections. Gender-based variations were also observed, with male drivers displaying a greater tendency to accept smaller gaps compared to female drivers. This indicates a higher level of risk-taking behavior among male drivers, which could be influenced by experience, confidence, or driving habits. Female drivers, on the other hand, were found to be more cautious, preferring to wait for larger gaps before making their move. These differences highlight the role of human factors in traffic decision-making and suggest that targeted awareness programs could help reduce high-risk driving behaviors.

Additionally, the speed of vehicles on the major road played a crucial role in influencing gap acceptance decisions. When the approaching vehicles were moving at higher speeds, drivers on the minor road became more cautious and tended to wait for larger gaps before attempting to cross. This behavior was particularly evident at intersections where visibility was limited, making it harder for drivers to accurately judge the speed and distance of oncoming vehicles. In contrast, at locations with better sight distance and lower traffic speeds, drivers were more likely to accept smaller gaps. Further analysis showed that traffic density and road conditions also had an impact on gap selection decisions. At intersections with higher traffic volumes, drivers displayed more cautious behavior, as frequent vehicle arrivals reduced the availability of safe gaps. Additionally, at locations where road markings were unclear or pavement conditions were poor, drivers hesitated longer before accepting a gap. This suggests that improving infrastructure and road signage could help reduce uncertainty and improve overall intersection safety.

Discussion

The results of this study emphasize the importance of integrating gap acceptance behavior into intersection design and traffic management strategies to improve road safety. The findings indicate that motorcyclists have a lower gap acceptance threshold, making them more vulnerable to accidents in mixed-traffic environments. Since motorcycles and three-wheelers operate alongside larger vehicles with different maneuvering characteristics, they often engage in riskier decision-making, leading to a higher likelihood of collisions. Addressing this issue requires a multi-faceted approach, combining engineering solutions, policy regulations, and awareness programs to create safer conditions at uncontrolled intersections. One of the most effective ways to enhance safety is through traffic management interventions. Measures such as reducing speed limits on major roads near intersections, introducing traffic calming measures, and improving lane discipline can help regulate traffic flow and make it easier for drivers to assess gaps more accurately. Additionally, clear and well-maintained road markings can assist drivers in making more informed decisions about when it is safe to cross or merge. Many intersections in urban areas lack proper signage and pavement markings, which can lead to confusion and hesitation among drivers. Implementing standardized markings and regulatory signs at unsignalized intersections can improve visibility and guide road users in selecting safer gaps.

Beyond infrastructure improvements, driver education and awareness campaigns play a crucial role in addressing unsafe gap acceptance behavior. Many motorcyclists and three-wheeler drivers may not fully recognize the risks associated with accepting smaller gaps, particularly when interacting with larger and faster-moving vehicles. Public safety campaigns, targeted at vulnerable road users, can help create awareness about the dangers of misjudging gaps and reinforce defensive driving techniques. Encouraging safer driving habits through training programs and road safety workshops could lead to more cautious and responsible driving behavior. From an engineering perspective, improving intersection geometry can significantly enhance safety at unsignalized intersections. Adjustments such as widening intersection approaches, increasing sight distance, and adding pedestrian islands can help reduce conflicts and improve visibility for all road users. The study's findings suggest that at locations where visibility was limited, drivers tended to wait for larger gaps, whereas intersections with better sight distance encouraged quicker but still cautious decision-making. This highlights the need for better intersection design standards that consider the unique challenges posed by mixed-traffic conditions in developing regions like Sri Lanka. Additionally, these research findings can be applied to microscopic traffic simulation models, which are widely used in transportation planning and policymaking. By incorporating real-world behavioral data into simulation software such as VISSIM or SUMO, traffic engineers can model different intersection scenarios and predict how modifications in infrastructure or traffic regulations might affect driver behavior. Such models can assist in optimizing intersection layouts, testing new safety measures before implementation, and improving overall traffic efficiency.

Conclusion

This research highlights the significance of gap acceptance behavior in intersection safety and provides valuable insights into the decision-making patterns of motorcyclists and three-wheeler drivers at unsignalized T-junctions in Sri Lanka. The study identifies critical gap values and explores the influence of vehicle type, gender, and speed on driver behavior. The findings suggest that lower gap acceptance values among motorcyclists increase their exposure to potential conflicts, highlighting the need for tailored safety interventions. To improve intersection safety, a combination of infrastructure enhancements, traffic management policies, and behavioral awareness programs should be implemented. Future research should explore real-time monitoring technologies and AI-based predictive models to develop more precise safety solutions. By integrating behavioral insights into transportation planning, road safety can be significantly enhanced, ensuring safer intersections for all users.

References Ashalatha, R., & Chandra, S. (2011). Critical gap through clearing behavior of drivers at unsignalised intersections. KSCE Journal of Civil Engineering, 15(8), 1427– 1434. https://doi.org/10.1007/s12205-011-1392-5 Bhatt, K., & Shah, J. (2022). Driver’s Risk Compelling Behavior for Crossing Conflict Area at Three-Legged Uncontrolled Intersection. In Intelligent Infrastructure in Transportation and Management: Proceedings of i-TRAM 2021 (pp. 39-52). Springer Singapore. Sandaruwan, A., Karunarathne, T., & Wickramasinghe, V. (2019). Estimating critical gap at roundabout under mixed traffic conditions in Sri Lanka. Journal of the Eastern Asia Society for Transportation Studies, 13, 80-92. Vajeeran, A., & De Silva, G. (2020). Identification of Effective Intersection Control Strategies During Peak Hours. Transportation Research Procedia, 48, 687-697.

ABSTRACT. City-wide traffic flow prediction is essential for Intelligent Transportation Systems (ITS) and provides the basis for traffic management and environmental monitoring. Despite advances in traffic modelling using simulation, statistics, and machine learning, in practice, predicting traffic flow for entire cities remains a challenge due to the natural spatiotemporal complexity of the problem and inconsistent input data availability. A key contribution of this research is a practical and reproducible workflow that researchers and practitioners can adapt and improve for different use cases in city-wide traffic modelling. Recent developments - such as opportunistic data sources supplementing sparse traffic counting sensors (Mahajan et al., 2022), graph neural networks (GNNs) emerging as state-of-the-art for spatiotemporal modelling (Jiang and Luo, 2022), and transfer learning for generalising pre-trained models to domains with limited data (Pan and Yang, 2010) - create new opportunities for research on model generalisation across road links and networks. Therefore, we propose a three-step-workflow consisting of (1) data preparation, (2) training a GNN for traffic flow prediction on motorways and arterials in a data-rich urban network, and (3) transferring the pre-trained model to a second network with minimal historical data. We demonstrate this approach using Paris as a data-rich network and Copenhagen as a data-scarce one, using available subsets of historical data from stationary sensors, speed data from floating car data, and aggregated simulation results. The models are benchmarked for different levels of data scarcity, and impacts on model accuracy, robustness, and transferability are discussed.

ABSTRACT. In recent years, the debate on a real diffusion of driverless autonomous vehicles has been growing, also thanks to the advancement of technologies that led to the creation of increasingly efficient prototypes. In this respect, while driverless transit is a reality in the case of railway fixed guideway systems, a few steps have been taken to implement automated road transit (i.e., buses). Tests have been conducted during the last decade, but most of them show that there is still no structured knowledge of the acceptance of users regarding this type of system. In particular, in Italy, some pilot projects have been carried out, but they have never resulted in surveys that characterize the acceptance of the system by the population, analyzing its distinctive features compared to other international practices. Nowadays with the advancement of low-cost technology, brain-computer interfaces have been widely used to control machines using electroencephalogram (EEG) signals. They are used as beneficial devices to develop biofeedback systems by exploiting the acquired signal. This paper presents the results of a survey conducted on a real test experience of an automated shuttle in the Italian city of Bari. The aim is to investigate the acceptance of the system and analyze its relationship with the population’s socio-demographic characteristics and current mobility habits. In addition, participants’ mental state about the experience by exploiting BCIs was analyzed with focus, relaxation, and stress level metrics, while emotion polarity was computed with EEG connectivity graphs metrics. Results provide first insights into the willingness to use the system but also pave the way for more in-depth investigation.

ABSTRACT. In the past decades, cities worldwide have experienced an increase in urban population and a change in shopping behaviors. More consumers are buying goods and services online, which has led to an increase in commercial vehicle traffic in urban areas. To cope with the increase in road and parking congestion, urban planners adopted different approaches to managing commercial vehicles. These approaches often lack the behavioral foundation of an understanding of how delivery drivers use the urban infrastructure to route and park their vehicles. This paper performs a comparative analysis of delivery drivers’ routing and parking behaviors across four delivery companies operating in two different geographical contexts. Data from more than 2316 deliveries and 1342 parking events that took place in Seattle (WA, US) and Stockholm (SE) are analyzed, and their routing and parking behaviors are compared. The project objective is to develop a data-driven science of urban freight route behaviors, supporting behavioral-informed strategies to manage commercial vehicle traffic in cities.

ABSTRACT. Recent studies reveal that public opinion shapes the uptake of urban air mobility, notably vertical take-off and landing (VTOL) aircraft, based on social, cultural, and political contexts. While traditional surveys have limitations disentangling such combined effects, social media could prove to be a valuable data source for tracking sentiment shifts across diverse regions. The current study examines how VTOL-related sentiments have evolved over time and across countries by analysing geo-tagged Twitter (X) data from 2010 to 2021. Leveraging Large Language Models (LLMs), Machine Learning (ML) models, and topic modelling techniques, it uncovers key themes and regional variations, offering deeper insights into the factors driving public discourse on VTOL technology.

ABSTRACT. A 2+1 highway represents a compromise between two-lane highways, which may or may not include passing lanes, and four-lane highways. However, the adoption of the 2+1 design remains limited in Brazil. This paper aims to evaluate the impact of this configuration in Brazilian conditions. The study gathered field data from a Brazilian two-lane rural highway to validate a traffic simulator previously calibrated for local two-lane highways with passing lanes. Subsequently, traffic simulations were performed for various hypothetical scenarios. Based on the simulated results, the 7th edition of the Highway Capacity Manual (HCM-7) was recalibrated to consider 2+1 highways in the country. The follower density values obtained using the revised HCM-7 method were closer to field data than those derived from the original, unmodified HCM-7.

ABSTRACT. The present study proposes to couple the BIDIM-GSOM (Bi-dimensional model –Generic Second Order Model) integrated traffic model of a very large-scale traffic network (Heni, et al., 2022) with a pollutant dispersion model to estimate emissions of key pollutants generated by traffic, such as CO, NOx, and CO2. The proposed methodology begins by determining the fundamental variables of traffic. Subsequently, vehicle unit emissions are estimated using a macroscopic emission model based on the average traffic speed, as a function of emission factors determined based on specific characteristics of the studied flow (Copert) (Heni, et al., 2023). Finally, a Gaussian dispersion model is implemented to determine pollutant concentrations at a significant urban scale in Marne-la-Vallée and on the A4 highway in France. The results are heavily influenced by input data. This study has proven useful in providing an estimation of concentration pollutant.

ABSTRACT. Background and Motivation A lot of interest has been shown towards automated vehicles and their potential impacts on traffic flow and emissions. However, most studies have focused on specific scenarios, often with optimistic assumptions unrealistic in the near term. Further, a wide range of assumptions and models has been used, making it difficult to form an overall picture of impacts. Impacts have rarely been studied separately for different vehicle types (Aittoniemi et al. 2024), although it is not clear whether impacts will be evenly distributed. HiDrive (n.d.), a European flagship project led by the car industry, aims to advance driving automation. With advanced technology enablers such as enhanced connectivity and positioning support, it seeks to reduce the need for human takeovers and enable use of advanced automated driving functions (ADFs) in urban and motorway settings. One part of the project is evaluation, estimating the societal effects of the ADFs and their enablers after market introduction, compared to current traffic. Objectives and scope To assess the impact potential of the enablers, it is necessary to first understand the impacts of automated driving without them. This study examines the impacts of automated, privately owned passenger cars without connectivity or other specific enablers, compared to the baseline of traffic today where automation is not widely implemented beyond advanced driver assistance systems (ADAS). Effects are first studied for typical traffic conditions on European urban roads and motorways, followed by an assessment of their large-scale significance across Europe. Specifically, the impacts on travel times, tractive energy use, and CO2 emissions are studied for different vehicle types, including users, non-users, and heavy-duty vehicles, for a time period of one year. The study focuses on impacts arising from the different driving behaviour caused by replacing human drivers with driving automation. It is acknowledged that secondary impacts such as changes in demand for car travel or accident-induced congestion are also relevant when considering impacts of driving automation, but the focus is set here specifically on isolating the direct impacts. Methodology Overall approach The methodology consists of three main steps: defining relevant traffic scenarios, estimating effect sizes per traffic scenario, and scaling up results to form an estimate of overall impacts on the European level, defined as the 27 EU member states as of 2023. The operational design domain (ODD) of the ADFs needs to be considered when estimating their potential impacts, and the ODD requirements have been defined together with the ADF developers. For motorways, it is estimated that the network fulfilled the infrastructural requirements, leaving weather as the only relevant dimension, as bad weather conditions such as heavy rain were excluded. For urban areas, ODD requirements pose more constraints. Based on high-level assessment an assumption has been made that roads classified as primary, secondary or tertiary in OpenStreetMap fulfil, on average, the infrastructural requirements of the ODD, including presence of distinct lanes and sufficient width of lanes. The weather dimension is relevant also for urban roads. Defining Traffic scenarios Traffic scenarios are used to cover systematically the main variations in road infrastructure and traffic demand, such that, as a whole, they represent the European motorway and urban road networks as well as possible. To find the most relevant conditions, data on European urban roads and motorways has been collected from OpenStreetMap, and traffic data from various open sources, including national contact points of EU member states and city websites. The level of detail of the scenarios needs to be balanced between capturing key dimensions while keeping the total number of scenarios manageable. In addition, availability of data on the European level poses constraints. The speed limit, number of lanes, the type of intersections, traffic volume and ADF penetration rate among passenger cars were defined as the main dimensions. In addition to passenger cars, heavy-duty vehicles are included in the simulations. For motorways, the whole European network is considered, whereas for urban roads the scope needs to be limited due to the higher levels of complexity and poorer data availability. In the first phase of analysis, the ~500 cities in the Urban Atlas (EU 2025) dataset are included. Estimating effect sizes with traffic simulations The microscopic simulation tool PTV Vissim is used to study impacts of the ADFs arising from different driving behaviour. Travel time impacts can be derived directly from simulation results, whereas energy use and emissions impacts are estimated with suitable tools using trajectory data from the simulations. Effects are defined as changes in the average travel time, energy use and CO2 emissions relative to vehicle kilometres travelled (VKT), per vehicle type, comparing the baseline (only human-driven vehicles) and treatment (a share of human-driven passenger cars replaced by automated vehicles) scenarios. The ADF behaviour was implemented with an external driver model. For human-driven vehicles, the Wiedemann99 (for motorways) and Wiedemann74 (for urban areas) models are used. The main differences compared to human-driven vehicles are longer desired time gaps (1.6 s compared to the average 1.05 s for human drivers), adherence to the speed limit, more stable acceleration patterns and less cooperative lane change behaviour. In addition, driving behaviour of all automated vehicles is assumed homogeneous, contrasting with the stochasticity of human driver parameters. The simulated vehicle fleet represents the average European passenger cars and heavy-duty vehicles. Scaling up impacts to European level Scaling up refers to aggregating results from traffic scenario simulations to the whole European motorway network and the main urban roads, utilising available statistics and data. The method for scaling up, developed earlier for motorways (Bjorvatn et al. 2021) will be used and expanded for urban roads. The process involves combining aggregated traffic and map data with the corresponding effect sizes from simulations. First, the share of VKT within and outside the operational design domain (ODDs) of the ADF is estimated for each traffic scenario. These VKT estimates are then combined with the previously derived effect sizes per VKT to produce an overall estimate of the direct impacts of the ADF across Europe. As weather conditions are local, the scaling up process uses the European NUTS3 classification of areas. In a final step, the estimates will be adjusted to account for potential indirect impacts, such as those resulting from changes in travel behaviour (total VKT) or traffic safety. Results and Discussion At the time of submitting this abstract, results from simulations are not available, as the work is currently ongoing. Results will be included in the full version of the paper.

ABSTRACT. The implementation of automation on public transport might mitigate the unsustainable effects associated with automated private vehicles, such as number of vehicles and increased trip length, while simultaneously enhancing safety, decreasing energy consumption, and improving mobility availability. Integrating Automated Buses (AB) into a transportation system could enable cities to establish a more efficient, environmentally sustainable, and accessible public transport network. This study investigates the influence on mode choice resulting from the introduction of ABs in urban environments. A Discrete Choice Experiments (DCE) was developed to explore consumer preferences for different transport modes (car, robo-taxi, bus or bicycle) and willingness to pay (WTP) for attributes such as travel time, service frequency, staff presence and role, autonomous driving capability, electric vehicles, and on-demand services. Highlighting the relative importance individuals assign to different bus attributes will offer guidance to the transport sector in their AB deployment strategies. A total of 2,004 responses were collected through a response panel using the online survey platform SurveyEngine (2024) between 14th May and 19th July 2024. The distribution process included enforcing quota constraints for age, gender, and household income to ensure that the sample was representative of the Scottish population. The questionnaire received ethical and governance approval from Edinburgh Napier University. The questionnaire was composed of a wide range of largely closed-ended questions such as socio-demographic characteristics, travel behaviour, and attitude towards the environment, technology, and vehicle automation. A key part of the survey was a discrete-choice experiment (DCE) where respondents were presented with four transport alternatives (bus, robo-taxi, car, and bicycle), each with different attribute levels related to travel time, walk time, wait time, cost (fares and fuel), parking charges, staff presence and role, and electric vehicles. Participants were given the context of the choice (e.g., regular trip, in good weather, travelling alone and arriving at their destination by 9am) before being presented with the choice sets. The software package Ngene (ChoiceMetrics 2021) was used to generate an orthogonal experimental design for the initial pilot study. The coefficients and their corresponding standard errors obtained from the pilot were used as priors to generate the final choice sets through a Bayesian efficient design using the Apollo package in R software (Hess and Palma 2019). The final design consisted of seven questions per respondent, distributed across three blocks to avoid cognitive overload for participants, while ensuring sufficient statistical accuracy (Szinay et al. 2021). The data from the DCE was analysed using a Multinomial Logit (MNL) model, which is suitable for situations where individuals are asked to choose from more than two alternatives (Hensher et al. 2015). The model was used to estimate respondents' willingness to pay (WTP) for different bus service attributes which provides a standardised measure of WTP, allowing comparisons across different service features and enabling the assessment of the relative value respondents place on each attribute. A total of 14,190 responses were received from 2,004 individuals. The car was selected most often (34%) followed by bus (25.1%) bike (23.3%) then taxi (17.6%). The alternative specific constants for bus, taxi and bike are negative and significant, indicating a baseline disutility for choosing these modes relative to the car. Increased costs (fares, fuel, parking) and travel times (in-vehicle, walking) are negative and significant across all modes. The absence of staff onboard buses is a strong negative factor influencing bus choice. Additionally, the presence of remote supervision for buses was associated with a further decrease in the propensity to choose the bus compared to scenarios where no staff were present. This suggest that participants may not perceive any advantages from remote supervision in the absence of staff. Possible explanations may include participants' incomplete understanding of the remote supervision concept as presented in the choice task, discomfort with being monitored remotely, or a belief that a fully automated system is more capable than one supervised remotely by a human. The MNL model also indicates that an electric AB fleet would significantly increase the proportion of individuals choosing the bus over alternative modes which reinforces the value of incorporating electric vehicles as part of AB trials. All attributes, except for bus wait-time and taxi remote supervision are statistically significant at the 1% level. The log likelihood comparison between the full model (including all attributes) and the reduced model (excluding bus wait time and taxi remote supervision) is not statistically significant. Therefore, including these two variables does not improve the model fit sufficiently to justify inclusion. Several explanations could account for the finding that bus wait time was not significant. The maximum bus wait time presented was 10 minutes, which was considered realistic due to the prevalent use of real-time information across Scotland, allowing travellers to better align their arrival at the bus stop with the bus's arrival. It is possible that the attribute level range was too narrow for respondents to perceive meaningful differences between the options. Alternatively, the result might show attribute non-attendance or dominance of other attributes. Finally, the non-significance may be due to a high level of variation among respondents’ preferences, which requires further investigation. A consideration of respondents WTP for different bus service features revealed that having a bus driver onboard who is responsible for all driving tasks showed the highest WTP at £4.98. Conversely, the WTP for remote supervision was negative at -£0.67, suggesting that respondents would not value any additional costs associated with this feature. Respondents showed a considerably higher WTP for a bus driver than for a bus captain, whose role is limited to customer care. This highlights the perceived higher value of having a bus driver on board, where a human is responsible for all driving tasks. Although operating ABs without an employee could yield cost savings, the current findings suggest that this approach may deter potential users in Scotland, increasing the likelihood of them opting for alternative transport options.

ABSTRACT. This study addresses a critical challenge faced by Multimodal Transport Operators (MTOs) in planning truck routes for containerized goods. By focusing on the integration of last-mile, first-mile, and road transport services, we propose an optimization approach to enhance logistics operations. Our method emphasizes strategic scheduling and coordination of deliveries and pickups, aiming to minimize travel costs, reduce empty trips, and improve resource utilization. The approach consists of two main steps: assessing a comprehensive transport order list to identify compatible service sets and employing a mathematical model to allocate specific transport tasks to available vehicles. Preliminary tests were conducted using a series of carefully designed test cases to evaluate the approach's performance and effectiveness. Key indicators were proposed too. The results indicate that the proposed optimization framework effectively enhances logistics efficiency and contributes to sustainable transport practices.

ABSTRACT. We propose the introduction of monetary incentives to reduce long bus dwell times at stops near demand hotspots. Passengers pay lower fares if they are alighting at a less busy stop. We suggest that an appropriate bus incentive can reducing bus dwell time. We designed a microscopic traffic scenario for the Kyoto City Bus line 201 with the SUMO simulation tool to test this. An intermodal routing tool was utilized to extract boarding and alighting times. We obtain the resulting passenger disutility including increased walking to their destination and compare it to reduction in bus operating costs. Additionally, initial results suggest that in-vehicle crowding is crucial for bus dwell time.

ABSTRACT. Transportation systems are pivotal in advancing sustainable urbanization and facilitating the transition to more environmentally conscious living practices. Over recent decades, European cities have experienced profound transformations, with suburbanization introducing new and complex challenges. Suburban areas, while often more affordable in terms of housing, are predominantly characterized by car-dependent urban designs. This reliance on private vehicles contributes to elevated commuting costs, reduced mobility, transport and time poverty, environmental pollution, increased accident rates, and diminished social interaction. To provide sustainable mobility solutions to urban populations, it is critical to comprehensively analyse their mobility patterns, personal characteristics, and the physical attributes of the built environment. This research focuses on the Barcelona Metropolitan Region, which benefits from a comprehensive public transport network, especially in certain areas. The primary objective of this study is to explore the interaction between travel behaviour, transportation mode choices, individual socioeconomic attributes, and residential location, utilizing detailed data sources. Specifically, the analysis leverages the annual travel surveys conducted in the Barcelona region from 2018 to 2023, in conjunction with land-use and other relevant datasets.

The findings of this study indicate that transportation policies have effectively fostered sustainable and prevalent mobility practices, particularly within the central areas of Barcelona. However, significant disparities persist between the inner city and suburban regions, where sustainable mobility adoption is markedly lower. The results underscore the need for further enhancements in transportation, urban, and land-use planning policies to promote equity and address the socioeconomic characteristics of residents.

ABSTRACT. Understanding how commuters choose their travel modes is essential for fostering sustainable mobility and reducing reliance on private vehicles. This study employs Multinomial Logit (MNL) and Neural Network (NN) models on survey data collected from employees in Rome, Italy, ensuring a fair comparison by applying identical preprocessing and evaluation metrics. While the NN model achieves slightly higher accuracy, statistical analysis indicates that the difference is not significant. The elasticity analysis highlights key factors shaping commuting choices, offering interpretable insights into travel behavior. These results demonstrate that, despite its lower accuracy, MNL provides strong predictive capabilities while preserving interpretability. This underscores the continued importance of traditional econometric models in transportation research, particularly for policymaking, where explainability is crucial.

ABSTRACT. Demand Responsive Transport (DRT) systems provide versatile transport operations and are capable of quickly adjust to fluctuating passenger demand. Unlike traditional public transport (PT), which operates with fixed routes and schedules, and with a pre-defined vehicle assignment, DRTs offer flexibility in vehicle paths, fleet sizes, and operating hours. This feature is an intrinsic characteristic of DRT systems and a key attribute in their design and associated decision-making processes. However, flexibility presents important design challenges, due to the multitude of potential configurations and the unique characteristics of each service area. While increased flexibility allows the system to adapt quickly to changes in demand and to provide more personalized services to passengers, it can also create some level of uncertainty around pick-up and drop-off times, potentially deterring time-sensitive passengers. As practice shows, the effectiveness of DRT configurations is heavily influenced by demand levels. Highly flexible operations are typically suited for low-demand areas, whereas higher demand may require reduced flexibility to maintain system efficiency. Furthermore, demand may grow to a point where the operators may question whether to continue operating as a DRT or shift to traditional regular public transport, with predefined routes and schedules, and a more efficient operation. This work specifically addresses the challenge of determining demand thresholds for DRT configurations and supporting the adoption of these systems instead of traditional PT. We have approached these problems first by reviewing existing literature on DRT design and then by using a simulation model to test various demand scenarios in a virtual environment, as a way to compare the performance and limitations of the different transport systems, offering insights into their operational trade-offs.

ABSTRACT. This study examines travel behaviours and modal choices among students, faculty, and staff across five universities in Italy, located in Milan, Naples, Cagliari, Padua, and Bari. The research focuses on understanding the factors influencing modal choices in distinct territorial and transport contexts, analysing key variables such as travel times, costs, and access, while considering the socioeconomic and professional differences within the academic community. Specifically, the study compares travel patterns of students and faculty/staff, using different specifications of Multinomial Logit (MNL) model to account for heterogeneous preferences and the willingness to pay for various aspects of the journey. Initial results highlight the differences in perceptions between students and faculty/staff, showing how these are strongly influenced by the specific context. The study’s large and diverse sample enhances the generalizability of the findings, offering insights for the design of targeted sustainable mobility policies and the prediction of their effects on demand.

Introduction Academic communities, including students, faculty, and administrative staff, play a crucial role in urban mobility due to the significant volume of trips generated by university campuses or decentralized campuses (Balsas, 2003; Rotaris & Danielis, 2014; Sottile et al., 2021). The increasing focus on sustainable mobility in academic institutions (Balsas, 2003; Shannon et al., 2006; Hancock & Nuttman, 2014), combined with the non-standard travel pattern within this population (Whalen et al., 2013; Hamad et al., 2021), makes universities an ideal setting for evaluating policies aimed at promoting sustainable commuting (Delmelle & Delmelle, 2012; Sottile et al., 2021). To develop effective policies, also in the perspective of viewing mobility as a service and proposing integrated and beneficial mobility solutions, it is essential to first identify the key factors influencing commute mode choices among academic populations (Rotaris & Danielis, 2014; Hamad et al., 2021; Sottile et al., 2022). Given the socioeconomic, professional, and travel behaviour differences between students and faculty/staff, cross-group comparisons are particularly valuable for understanding how different segments of the academic community may respond to different policy measures (Shannon et al., 2006; Romanowska et al., 2019). Most studies primarily examine student travel patterns within single institutions, with fewer exploring differences across academic roles (Shannon et al., 2006; Akar et al., 2012; Hamad et al., 2021) or conducting multi-university comparisons (Balsas, 2003; Davison et al., 2015). Moreover, studies that do compare multiple universities often rely on descriptive analyses, which may lack the statistical depth needed to account for variations in mode attributes and socioeconomic factors in addressing policy implications. This study addresses these gaps by conducting a comparative analysis of commute mode choices among students and faculty/staff at five Italian universities in five different cities (Milan, Naples, Cagliari, Padua and Bari) each representing distinct territorial and transportation contexts. By analysing key variables such as access/egress time, in-vehicle travel time, and travel costs while controlling for socioeconomic characteristics, this research provides insights into how policies may affect different academic groups. Furthermore, the cross-context approach allowed us to verify how the willingness to pay for various aspects of the journey differs across contexts, highlighting how it is closely related to the characteristics of the transport offer, since in this case the reference population is the same across the different contexts.

The survey The data used in this study were collected as part of a revealed preference (RP) survey designed to analyse the current travel habits of the academic population at five Italian universities: the University of Milan-Bicocca, the University of Naples-Federico II, the University of Cagliari, the University of Padua, and the Polytechnic of Bari. Data collection took place between December 2023 and March 2024 and consisted of an online survey distributed via mailing lists and in-person dissemination directly in classrooms. The survey collected data on participants’ university roles, campus locations, and mobility behaviour, including their usual mode of transport, frequency of attendance, and available alternatives. It also gathered socioeconomic and demographic information, including home addresses, which were used to estimate trip distance, travel time, and travel costs. The study contexts The five pilot study contexts differ in territorial characteristics but, above all, in their transport offerings. Milan and Naples are significantly larger cities, each serving as the capitals of the second and third most populous metropolitan areas in Italy, with populations around 3 million. In contrast, Cagliari, Padua and Bari are significantly smaller, with populations ranging from about 150,000 to 300,000. Public transportation varies widely across these cities. Milan has the most extensive system, with a comprehensive rail and metro network. Naples relies mainly on a metro system with three lines and regional rail, while Cagliari, Padua and Bari primarily depend on bus services, with limited rail options. Cycling infrastructure is notably more developed in Padua, where cycling is a more common mode of transport due to the city’s smaller size and historical cycling culture. Milan has also seen increased cycling usage due to recently expanded infrastructure and sharing mobility options. In contrast, bicycle use in Naples, Cagliari and Bari remains limited due to the terrain morphology and inadequate/non-existent cycling infrastructure, which makes cycling less accessible and less safe.

Data analysis A data cleaning process was carried out to remove incomplete or inconsistent responses, as well as non-georeferenced origins and destinations. Additionally, due to the low number of bike users in Naples, Cagliari and Bari, which did not allow for the use of this alternative in the models, it was decided to exclude these individuals from the dataset. Google API services were used to calculate the transport attributes of the chosen alternatives and to simulate those of the available but unchosen alternatives. The total sample consists of 13,002 valid observations (approximately 5.7% of the total population of enrolled students and faculty/staff) across five universities: Milan-Bicocca (2,365; 6.2%), Naples-Federico II (3,649; 4.7%), Cagliari (1,188; 4.6%), Padua (5,359; 7.0%) and Polytechnic of Bari (441; 4.1%). About two-thirds of the respondents are students and one-third faculty/staff, with Naples showing a higher student proportion mainly due to massive interception in the classroom, and Bari a higher faculty/staff presence. Milan-Bicocca has the longest commute distances, while Cagliari and Bari have the most trips under 5 km. Modal split reveals high public transit use in Milan and Naples, significant bicycle use in Padua, a high percentage of walking trips in Bari, and a greater reliance on cars in Cagliari and Bari.

Methodology Investigating the differences between contexts and in the perception of the Level of Service attributes of alternatives, as well as the willingness to pay, by students compared to faculty and staff, is the focus of this article. In this regard, we decided to calibrate Multinomial Logit (MNL) models for each context, incorporating a scaling factor in the utility specification associated with students that checks for different error variances in the subsample (students vs. faculty/staff). The scaling factor is inversely proportional to the error variance and allows for the same variance between the two subsamples, but it does not enable us to examine a difference in perceptions of travel attributes. Therefore, we test heterogeneous preferences through MNL models (one for each context) with interaction terms between role and key explanatory variables, such as travel time and cost. This approach also allowed us to account for differences in willingness to pay for various travel attributes and the different time values associated with various phases of the trips in the contexts and academic roles. Additionally, elasticities are calculated to highlight differences in the effectiveness of potential future policies aimed at increasing the share of sustainable travel options across academic roles and contexts.

Early Results The estimated scale factor for students is consistently greater than 1, indicating their choices are more deterministic and less influenced by unobserved factors compared to faculty/staff, whose scale factor is fixed at 1. The values are 1.21 (t-stat = 5.11) for Milan, 2.52 (t-stat = 4.89) for Naples, 1.60 (t-stat = 6.33) for Cagliari, and 1.38 (t-stat = 12.0) for Padua. In contrast, Bari stands out with a lower student scale factor of 0.52 (t-stat = 3.98), suggesting unobserved factors play a larger role in student choices. To explicitly test for preference heterogeneity, interaction terms are introduced. Early estimates focus on interaction terms for public transit (in-vehicle) travel time. The results consistently show that individuals place greater importance on access/egress and waiting times than on in-vehicle travel time. They also perceive higher disutility for active mobility and car travel times compared to public transportation, likely due to the physical effort required for active mobility and the stress associated with driving. In terms of interaction effects, students generally perceive (in-vehicle) transit travel time as less burdensome than faculty/staff. Notably, the Naples subsample shows the highest difference in perceived disutility for in-vehicle travel time on public transport compared to other contexts, with the Cagliari subsample showing no significant interaction effect. This could highlight unique differences in the perception of transit travel time in Naples, potentially linked to the fact that students have free access to public transportation, unlike in the other contexts. This difference is reflected in the value of travel time (VoT), with more significant discrepancies between academic groups in Naples than in other cities. Overall, the results indicate consistent VoT for private car use across contexts (approximately 16 €/hour, in line with other studies in Italy (Wardman et al., 2016)). However, the VoT for (in-vehicle) transit travel time is slightly higher in Naples and Cagliari (5 to 10 €/hour) compared to Milan, Padua, and Bari (3 to 5 €/hour). This pattern may be due to the better performance of public transport in Milan and Padua, as well as the presence of cycling as a viable alternative, which could reduce the willingness to pay for a reduction in in-vehicle transit time. In Bari, the significantly shorter average commute distance might have contributed to this result.

Conclusions The initial results clearly highlight the importance of analysing current travel behaviours in different contexts to shed light on how perceptions and modal choices change, not only depending on the role and associated characteristics, but also, and especially, in relation to territorial differences and the transport offer. Detailed information on modal choice, differentiated by context and role, enables the definition of targeted policies aimed at making academic mobility demand even more sustainable, particularly that of faculty and staff, which is generally less sustainable than that of students, and allows for accurate demand forecasting. The diversity of contexts also helps to highlight the impact of the presence of an extensive bike network on modal choice, with Padua being an example. Similarly, Milan serves as an example for public transportation, in terms of availability and the number of services, etc. A strength of this work lies in the large sample size across different contexts, which enhances the robustness as well as the generalizability of the results to other contexts with similar characteristics.

Acknowledgements This study was carried out within the MOST—Sustainable Mobility National Research Centre and received funding from the European Union Next-GenerationEU (PIANO NAZIONALE DI RIPRESA E RESILIENZA (PNRR)—MISSIONE 4 COMPONENTE 2, INVESTIMENTO 1.4—D.D. 1033 17 June 2022, CN00000023).

References Akar, G., Flynn, C., Namgung, M., 2012. Travel choices and links to transportation demand management. Transp. Res. Rec. 77–85. https://doi.org/10.3141/2319-09 Balsas, C.J.L., 2003. Sustainable transportation planning on college campuses. Transp. Policy 10, 35–49. https://doi.org/10.1016/S0967-070X(02)00028-8 Davison, L., Ahern, A., Hine, J., 2015. Travel, transport and energy implications of university-related student travel: A case study approach. Transp. Res. Part D Transp. Environ. 38, 27–40. https://doi.org/10.1016/j.trd.2015.04.028 Delmelle, E.M., Delmelle, E.C., 2012. Exploring spatio-temporal commuting patterns in a university environment. Transp. Policy 21, 1–9. https://doi.org/10.1016/j.tranpol.2011.12.007 Hamad, K., Htun, P.T.T., Obaid, L., 2021. Characterization of travel behavior at a university campus: A case study of Sharjah University City, UAE. Transp. Res. Interdiscip. Perspect. 12, 100488. https://doi.org/10.1016/j.trip.2021.100488 Hancock, L., Nuttman, S., 2014. Engaging higher education institutions in the challenge of sustainability: Sustainable transport as a catalyst for action. J. Clean. Prod. 62, 62–71. https://doi.org/10.1016/j.jclepro.2013.07.062 Romanowska, A., Okraszewska, R., Jamroz, K., 2019. A study of transport behaviour of academic communities. Sustain. 11. https://doi.org/10.3390/su11133519 Shannon, T., Giles-Corti, B., Pikora, T., Bulsara, M., Shilton, T., Bull, F., 2006. Active commuting in a university setting: Assessing commuting habits and potential for modal change. Transp. Policy 13, 240–253. https://doi.org/10.1016/j.tranpol.2005.11.002 Sottile, E., Giacchetti, T., Tuveri, G., Piras, F., Calli, D., Concas, V., Zamberlan, L., Meloni, I., Carrese, S., 2021. An innovative GPS smartphone based strategy for university mobility management: A case study at the University of RomaTre, Italy. Res. Transp. Econ. 85, 100926. https://doi.org/10.1016/j.retrec.2020.100926 Sottile, E., Tuveri, G., Piras, F., Meloni, I., 2022. Modelling commuting tours versus non-commuting tours for university students. A panel data analysis from different contexts. Transp. Policy 118, 56–67. https://doi.org/10.1016/j.tranpol.2021.12.019 Wardman, M., Chintakayala, V.P.K., de Jong, G., 2016. Values of travel time in Europe: Review and meta-analysis. Transp. Res. Part A Policy Pract. 94, 93–111. https://doi.org/10.1016/j.tra.2016.08.019

ABSTRACT. A resilient transportation system is critical to ensuring the mobility of people and goods, even under emergency conditions. It also helps reduce the economic and social costs associated with disasters and improves the quality of life for citizens. It is a complex concept involving several dimensions, including the redundancy of infrastructure, the diversity of transportation assets, the adaptability of the system to sudden changes, and its ability to withstand stresses. The relationship between resilience and concepts such as robustness, reliability, risk, and vulnerability needs to be explored further to get a more in-depth picture of the problem. In this context, this work proposes a comprehensive survey of methodologies for resilience assessment. The study explores the two main evaluation method categories: functionality-based metrics, which consider both structural and operational factors, and topological metrics, which prioritize network structure. This analysis also aims to clarify concepts and provide useful insights to researchers and practitioners for resilience-oriented transportation planning.

ABSTRACT. This research presents a dynamic fleet deployment model (DFDM) designed to improve urban mobility by addressing transit disruptions through real-time reallocation of shared micro-mobility resources. The study investigates the role of shared micro-mobility, such as bicycles, in alleviating the impacts of unpredicted disruptions in public transit systems. The DFDM integrates several components, including disruption detection, demand prediction, and dynamic fleet reallocation, utilizing machine learning techniques such as CNN-LSTM for disruption detection and stacked LSTM for demand prediction. The system's effectiveness is tested using a realistic disruption scenario in an example network. The framework optimizes fleet distribution through two cooperating optimization models, minimizing commuter delays and reallocation costs. A simulation environment in MATLAB evaluates the performance of the DFDM under different conditions, with key performance indicators focusing on delay reduction, fleet efficiency, and resource repositioning costs. This research enhances urban transportation resilience by integrating micro-mobility into transit recovery strategies.

ABSTRACT. Regions, cities but also medium-sized towns are regularly confronted with the necessity to assess the impact of transport strategies, action plans or transport infrastructure measures. Within the Sustainable Urban Mobility Planning (SUMP) process e.g., municipalities are invited to build and jointly assess scenarios depicting the development of the transport system and mobility behaviour as part of the strategy development process.

Simulation tools and their underlying transport models play a crucial role to support these processes, both on micro and macro level.  However, output data of these models are usually not suited for a comprehensive sustainability assessment. It was thus the aim of the research project "MOTUS - Mobility Transformation: Understanding key factors for sustainable and resilient transport" to develop a suitable assessment tool for this application. Based on a profound literature analysis, relevant sustainability impacts have been identified and combined with an impact evaluation framework allowing for an indicator-based assessment of scenarios. Indicators have been adapted to the input and output data commonly available in transport models. 

The sustainability assessment approach has been tested with an agent-based transport simulation, built for the city of Bad Hersfeld in Hesse, Germany, with the aim to explore strategies to improve the overall sustainability of the transport system.

ABSTRACT. (The extended abstract is attached as a PDF file with this submission.)

ABSTRACT. The planning processes of an airline are complicated due to the multitude of options. Additionally, it takes a long period of time from the first planning steps, like creating the network and defining the fleet, until the flights are executed. The development of the digital airline twin (DAT) at the German Aerospace Center (DLR) is motivated by the objective of simulating and optimizing these processes. The DAT will be a general tool which can be used for airlines with different business strategies. It represents not only all types of existing airlines but also possible future concepts and technologies. Part of the planning processes is assigning each individual aircraft to a sequence of specific flights within a given timeframe, known as aircraft rotation. This includes the necessity for regular maintenance for each aircraft in order to ensure operational efficiency and safety. Due to varying regulatory strategies within the legal framework that airlines may use to fulfill the maintenance requirements, different approaches are implemented to model the maintenance events. This paper describes these approaches and the general functionality of the model, which is based on a time-space network to represent the flight schedule and solved with Gurobi.

ABSTRACT. This study uses agent-based simulation to address the challenge of multimodal passenger access in airports during disruptions. We propose a system with two orchestrator agents: the first coordinating airside actors, while the second coordinating landside actors. We explore the impact of coordination measures, such as passenger rerouting and tactical flight delays, implemented by the orchestrators. A case study highlights the effectiveness of these measures in reducing missed flights and passenger delays.

ABSTRACT. This paper explores seat allocation and revenue management strategies in the airline industry. Its primary objective is to examine the impact of different reservation systems on total revenue of full-service airline by developing a simulation model of the reservation process for a non-stop flight. The analysis revealed that the non-nested model achieves better cabin occupancy, while the nested model generates higher revenue.

ABSTRACT. Inland Waterway Transport (IWT) is a cornerstone of the European Union’s sustainable mobility strategy, providing an eco-efficient alternative to road freight. However, its potential is constrained by uncertainties in waterway navigability and demand. This paper introduces an integrated tactical-operational decision making framework, which explicitly addresses these challenges through a coordinated process of tactical planning and operational decisions under uncertainty, reinforced as a bidirectional feedback mechanism. At the tactical level, seasonal service schedules are optimized using probabilistic water-level forecasts and demand projections to preemptively allocate resources (e.g., vessels, and terminals) while maintaining flexibility for operational adjustments. At the operational level, tactical plans (e.g., scheduled services) support operational decision-making. Revenue management principles guide demand acceptance, prioritizing high-value cargo, while dynamic re-routing algorithms adjust demand itineraries to mitigate disruptions like sudden water-level drops or demand surges. Furthermore, operational information—such as cargo bookings, demand itinerary deviations, and resource utilization—is systematically fed back to the tactical layer, enabling potential recalibration of schedules and capacity thresholds. By integrating tactical planning with operational adaptability, this study provides actionable insights to enhance the resilience, efficiency, and competitiveness of IWT within the European logistics network.

ABSTRACT. Urban Vehicle Access Regulations (UVARs) are policy measures aimed at regulating vehicle access in urban areas, taking into account criteria such as emission standards, vehicle classifications, and time-based restrictions. UVARs are crucial to enhancing environmental and traffic management objectives. While there have been relevant endeavours in digitising those regulations, the lack of harmonised quality definitions and assessment methods hinders the effective reuse of UVAR data by service providers. This paper addresses this gap by developing a Data Quality Framework (DQF) for UVARs, under the umbrella of the NAPCORE project. The framework was built by tailoring general data quality dimensions to UVAR-specific needs, defining a set of criteria and metrics, and introducing a method to compute quality indicators. It comprises seven quality dimensions, 12 criteria, and 15 metrics, and was tested through a case study using a real-world Low Emission Zone (LEZ) dataset from the city of Lisbon (Portugal). Despite some limitations, the framework offers a flexible approach for assessing and improving the quality of UVAR datasets, setting the ground for higher availability of UVAR data and facilitating its reuse and delivery to commuters, thereby contributing to urban mobility goals.

ABSTRACT. This study explores what makes people in Jakarta, Indonesia, want to use automated buses (ABs) using the extended Unified Theory of Acceptance and Use of Technology 2 (UTAUT-2) model. Indonesia has a lot of traffic accidents, and most are caused by human error. ABs could help make the roads safer. We wanted to investigate what people think about ABs, so we expanded by including the "environmental driving compliance" (EDC) factor in our model. EDC refers to how well people follow traffic rules. We surveyed 1,000 people in Jakarta, making sure we talked to different types of people. We then used an Exploratory Factor Analysis (EFA) method to group people's opinions into four main ideas: Perceived Personal and Social Value (PPSV), EDC, Trust in Policies (TP), and Risk Perception (RP). Using Structural Equation Modelling (SEM), we found that people who think ABs are useful, the roads are safe, and they trust the government are more likely to want to use them. If they're worried about the risks, they're less likely to do so. We also found that factors such as age, gender, income, education, tech skills, and how often people travel affect their feelings about ABs. Our findings can help policymakers in Jakarta come up with ways to get more people to use ABs. To do this, they need to show people how ABs are useful and safe and make sure everyone's concerns are addressed.

ABSTRACT. This study examines the adoption of Battery Electric Vehicles (BEVs) in Indonesia, analyzing the impact of convenience policy measures alongside behavioural and structural factors using Structural Equation Modeling (SEM). The findings indicate that convenience policy measures (CPM), such as traffic restrictions exemptions, dedicated parking and dedicated lanes, play a pivotal role in shaping behavioural intentions (BI), particularly when mediated by facilitating conditions (FC). Key predictors of BEV adoption include environmental concern (EC), personal innovativeness (PI), and price value (PV), while performance expectancy (PE) and social influence (SI) were insignificant. The study highlights the critical importance of enabling infrastructure and targeted policies in driving sustainable transportation behaviours. The study concludes that CPM can drive BEV adoption but depends on supportive FC. This highlights the critical importance of enabling infrastructure and targeted policies to drive sustainable transportation behaviours.