Abstracts Track 2024

Area 1 - Intelligent Transport Systems and Infrastructure

Nr: 83

Traffic Control Centres for the Optimization of Federated Networks in a CCAM Ecosystem


Franco Filippi, Facundo Storani, Roberta di Pace and Stefano de Luca

Abstract: We are currently witnessing a significant development and advancement in the field of connected, cooperative and automated mobility (CCAM). This progress is closely linked to the need for proactive traffic state estimation, forecasting, and management systems that harness the abundance of data originating from the CCAM context. While the technological landscape is well-established, there is a pressing need for major advances in two areas: 1) traffic flow modelling to effectively assess the impacts of connected and automated vehicles (CAVs) on transportation networks, and 2) traffic management and control strategies. Regarding the second point, operationally, the development of CCAM mobility occurs within the framework of a real ecosystem typically managed by traffic control centres. These centres exchange input-output data on multiple scales and involve various operators. Multi-scale modelling facilitates the discretization of the road network into several sub-networks, not only for analysis and control purposes but also for its management. This is evident in the representation of road networks as ‘federated’ networks, with each sub-network managed by a different operator. The research presents the preliminary major findings of the Project “Connected Networks and Smart Infrastructure” within the Italian National Centre for Sustainable Mobility, particularly focusing on the field of Optimization Networks in the CCAM ecosystem. The focus is on the specification of traffic control centres capable of managing enhanced “federated” networks in mixed traffic flow conditions, encompassing both human-driven and connected vehicles.

Area 2 - Intelligent Vehicle Technologies

Nr: 78

Sequential Bidding for Merging of Autonomous Vehicles.


Dmitrii Tikhonenko, Mihalis Markakis and Kalyan Talluri

Abstract: The recent advent of autonomous driving and 5G-based inter-vehicle communication technologies create enormous opportunities for fair and efficient management of vehicular traffic. It offers a chance to reconsider some of the basic principles of traffic management in situations such as merging and intersection management. In a broader sense, in congested environments where customers have heterogeneous time valuations, there is typically a tradeoff between fairness and efficiency, with the “First In, First Out” discipline on one end of the spectrum and priority queueing on the other. The quest for breaking this fairness-efficiency tradeoff has prompted the study of mechanisms through which agents with private time valuations can trade positions in queues. There is a notable body of literature on virtual queues, e.g., call centres, where the queue is unobservable to customers and, hence, each of them submits only a single bid. In this paper, we investigate the design of such mechanisms for queues that are observable, and dynamic bidding becomes a possibility. Concretely, we study a single driver on the blocked lane of a two-lane highway that must merge into a dense platoon of vehicles on the free lane; effectively, a customer joins a single-server queue. We focus on sequential bidding mechanisms, and we analyze the performance of two diametric opposites: a mechanism that bids from tail to head of the platoon (T2H) and one that bids from head to tail (H2T); the former imposes no negative externalities on free-lane drivers, hence is geared towards social welfare, whereas the latter clearly favors the merging driver. In both cases, we show that optimal bids and value functions can be computed via simple Dynamic Programming recursions. Moreover, for uniformly distributed time valuations, we prove that the expected social welfare of T2H is close to a partial-information social optimum and that the expected social welfare of H2T is lower than that of T2H, as long as the platoon is not too short. Finally, we propose fast Approximate Dynamic Programming algorithms, which lead to bids that are very close to the optimal ones, making them suitable for practical implementation. Notably, such mechanisms can be extended to other transportation areas, such as airway and canal usage. While promising, the T2H mechanism requires a notable level of enforcement upon traffic agents, as well as developing lower-level specific communication protocols and rules.

Nr: 79

Computer Vision Approaches for Vehicle Sideslip Angle Estimation


Basilio Lenzo, Leonardo Serena, Mattia Bruschetta and Ricardo de Castro

Abstract: Vehicle sideslip angle, defined as the angle between the longitudinal axis of a vehicle and its velocity vector, is a very important quantity for automotive control, specifically for vehicle stability control and safety. Unfortunately vehicle sideslip angle is very hard to access directly, therefore a variety of estimation methods have been developed so far. Such estimation methods are essentially based on model-based approaches or neural networks. Yet, essentially, there still is no reliable method to estimate sideslip angle. On the other hand, dedicated sensors exist but they are bulky and very expensive (> 20000 Eur), i.e. unpractical for application on standard passenger vehicles. This work looks at the problem from a fresh angle, by investigating possible solutions to measure vehicle sideslip angle via computer vision techniques, harnessing recent improvements in computer vision algorithms. Preliminary experiments on a radio-controlled scaled vehicle showed promising results using the "phase correlation" algorithm. They were presented in the our IEEE Metrology paper published in 2023. Since then, the project moved forward and we were able to test this methodology on two full-scale vehicles, i.e. a FIAT 124 Spider and an Alfa Romeo Giulia, along several maneuvers (e.g. double lane change/obstacle avoidance, drifting, skidpad etc.) with very promising results.

Nr: 82

Exploring Barriers to Users’ Adoption of a Traffic Risk Prediction and Warning App


Henrik Lechte, Eileen Vattheuer, Jannes D. Menck and Katharina Mengel

Abstract: Traffic risk prediction and warning applications promise to enhance road safety. However, some applications might face challenges based on special characteristics such as potential distractions to drivers, continuous data collection, and data networks effects. In this context, this study investigates factors influencing the adoption of a smartphone application that provides data-driven warnings for traffic hazards, while also relying on users to contribute their driving data to generate these warnings. Employing the innovation-decision process within the context of Rogers' diffusion of innovations theory as a high-level framework, this small-scale, exploratory study systematically organizes user perceptions gathered from semi-structured interviews. Key concerns include aspects such as privacy and data security, doubts about prediction accuracy, perceived lack of personal relevance, and the possibility of integration with other applications such as navigation tools. This study underscores the potential for broader research to further explore and validate the findings, enriching our understanding of potential adoption barriers for traffic safety applications. Gaining insight into these barriers can inform application developers and policymakers in addressing users’ concerns, promoting wider adoption, and enhancing road safety.

Area 3 - Data Analytics

Nr: 81

Information System Design for Utility Service Fleet Operations: The Use of Pooled Vehicle Trajectory Data for Super-Additive Analytics


Mathias Willnat, Christoph Prinz and Lutz Kolbe

Abstract: Urban traffic comprises not only individual private mobility and commercial freight transport but also a wide range of utility services serving as critical and social infrastructure. These services include city and regional bus services, fire brigade, emergency ambulance, patient transport services, social aid services, regular garbage collection, bulk waste collection, street cleaning, sewer cleaning, snow removal and police/security services. These services are, at least in Europe, typically operated or commissioned by public sector organisations. A robust and low-interference road network, along with tailored operations, are crucial prerequisites for efficiently providing these services. In this context, the utilization of motion data represents an essential approach to uncover and address inefficiencies within the infrastructure or operations. However, in our study area, which is a medium-sized German city, many of these vehicle movements are not collected as trajectory data by the public organisations that operate them. Some of the operators that do technically record the vehicle movements do not have adequate data interface access or the personnel capacity to process and analyse the data sets. Therefore, not only is the opportunity to enhance their own operations left untapped, but planning bodies in public administrations also have at best limited access to the trajectories of the transport services they commission and (co-)fund. This creates a hindrance in identifying neuralgic traffic points, e.g. due to recurring bottlenecks that affect the efficiency of public utility services in particular and urban transport in general. This contribution reports on a research endeavour to address the issue. Firstly, several public vehicle fleets have been equipped with position tracking devices. A harmonised database is being piloted, which also draws on external data sources (in particular geodata, weather data, OEM data on private car movements). Secondly, participatory workshops are being held with public stakeholders to develop use cases for possible data analyses. The aim is to explore how operational, planning, and strategic questions can be answered. Finally, a typical interactive tool will be made available to stakeholders, addressing the identified use cases with appropriate data analytics. We pool trajectory data from fleets of different vehicle types and service profiles to synergistically exploit enhanced evaluation potential. Conceptually, this is referred to as super-additivity, which is a synergistic relation where the value resulting from the combination of two segments is greater than the sum of their individual values. Our research is guided by the paradigm of Design Science Research. In this contribution, we discuss the contexts, hurdles and results of our research. In particular, we provide structured overviews of the context dependencies that emerged from the qualitative investigations of this study (Relevance Cycle), a categorisation of approaches to vehicle trajectory-based traffic obstacle analysis identified in the literature (Rigor Cycle), and identified design principles that emerged as outcomes from our conduct. Acknowledgements: This research received support from mFUND, a financial assistance programme sponsored by the German Federal Ministry of Transport and Digital Infrastructure.

Area 4 - Smart Mobility and Sustainable Transport Services

Nr: 80

Transfer Penalties in Active Transport Last Mile Connections: Evidence from a Stated Choice Experiment


Mathias Willnat, Felix Kegel, Christoph Prinz and Lutz Kolbe

Abstract: Transport planners and policy makers are increasingly seeking to reduce private car traffic in urban areas and promote more sustainable modes of transport. Increasingly, urban mobility policies include designated multimodal mobility hubs. This way, multimodal linking of diverse transport modes aims to increase overall connectivity in urban areas while facilitating more sustainable mobility. Naturally, the success of multimodal transport concepts is partly dependent on the perceived attractiveness of the vehicle transfers involved. Existing literature indicates that transfers are generally perceived as disadvantageous by passengers. Several studies provide insights into the perceived utility loss due to transfers (transfer penalties). However, most scenarios examined focus on transfers between vehicles of mass transit, such as subways and buses. Our study investigates transfer penalties in the context of transfers from bicycling to mass transit to extend this knowledge to include first-mile active transportation, as micro-mobility is often considered the priority solution in first/last-mile mobility concepts. We report on the design and analysis of a stated-choice experiment with over 8,500 individual choices. The experiment required participants to choose one of four options for a work commute from a visualization in the design of a route planning app: (a) a slow train connection, (b) a bicycle from a bikesharing service, (c) a combination of bikesharing and a faster train connection (allowing time savings at the cost of the inconvenience of changing modes), and (d) a private car connection that is artificially expensive due to a city tax scenario. We varied costs and trip durations across the instances. We tested the effect of visual design clues for crowding, bicycle availability and battery powered pedal support of shared bicycles. We present the results of our mixed logit model regressions. In addition to quantifying the effects of different influencing factors, we also quantify the transfer penalty. Academia typically reports three components into which transfer penalties can be decomposed: (1) the walking time associated with a transfer, (2) the waiting time associated with a transfer, and (3) a pure transfer penalty referring to any additional disutility induced by factors such as individual transfer uncertainty perceptions, additional mental effort and activity disruption, and the perceived reliability risk of the modes involved. Depending on the model specifications, the pure transfer penalty alone equates to about 7 to 10 minutes of train travel time. Adding walking and waiting times, single transfer journeys that include transit and bikesharing hence are seen as unattractive as journeys that are at least 10 minutes longer but do not include a transfer, making this an unviable option in many urban short-distance contexts. We argue that the disutility caused by transfers is largely under-recognised in academia, public debate and planning practice, where connectivity is often understood in terms of travel time alone. Our results may provide one element of an explanation as to why many concepts for facilitating first/last mile connectivity through improved alternative options (pull factor strategies), such as mobility hubs, park and ride stations and public bikesharing systems, often exhibit a disappointingly slow uptake; especially when not accompanied by push measures directed at unwelcome forms of mobility.