- 08:30-09:15 - Registration, welcome coffee & breakfast
- 09:15-09:30 - Opening speech
- 09:30-11:00 - Session 1 - Road Network Modelling and Control
- 1. Ludovic Leclercq (IFSTTAR ), “Trip-based MFD approach for modelling large cities and first application to parking” ➕
This talk focuses on modelling traffic dynamics at large-scale city scale based on the Macroscopic fundamental diagram (MFD). In particular, a recent trip-based approach is proposed: within a road network that defines a urban area all vehicles speed are driven by a simple behavioral rule, the MFD, while each vehicle travel distances are individualized. This formulation overcomes some limitations of the classical accumulation-based MFD model for simulation purpose. A simple and very efficient event-based numerical scheme for the trip-based approach is developed for the single reservoir problem. This scheme can be extended to tackle the multi-reservoir problem by properly defining how boundary conditions should be applied at the reservoir perimeters. Finally, we will exemplify this approach by addressing the problem of dynamic on-street parking search. This latter phenomenon can easily be represented in the trip-based framework as travel distance can be dynamically adjusted to the parking occupancy. the capabilities of the full framework is illustrated based on three different scenarios. The first two correspond to strategies with static and dynamic (reactive) switch of the demand from on- to off-street parking. The third scenario assesses the effect of a smart-parking technology that informs the users when a free parking spot is available on one of the downstream links at each intersection. In such a case, the model permits to estimate the benefit for the equipped users but also the impacts on all other vehicle categories.
- 2. Guido Gentile (University of Roma 1 ), “Trajectory based vs Shock-wave based solutions of the General Link Transmission Model” ➕
- 3. Andy Chow (University College London ), “Control strategies for network efficiency and resilience with route choice” ➕
This study formulates and analyses different urban traffic control systems with consideration of real time variations and uncertainties in traffic flow and travel behaviour. The control systems range from centralised, decentralised, to distributed for urban network traffic control. In a centralised system, all traffic information will be sent to a central controller or agent which is responsible for deriving and implementing all control actions. Based upon the store-and-forward model, we present the linear quadratic TUC control system as the representative which aims to minimise total network queues. We then demonstrate a decentralisation of the original TUC formulation through the alternative direction method of multipliers (ADMM). We compare the performance of this decentralised version of TUC with respect to the original centralised one over different network and network settings. For the distributed counterparts, we consider the recently established max-pressure (MP) control strategy. As a distributed system, the MP controller does not require knowledge of global network inflow and it adjusts local signal settings based upon local queue length measurements. The control systems are implemented on the SUMO simulation platform, incorporated with a dynamic routing algorithm, over different network topologies including one-dimensional arterial and two-dimensional grid networks. It is found that decentralised / distributed systems could indeed outperform their centralised counterparts when there is freedom (e.g. route choice) allowed in the network. This study contributes to the field of traffic control design.
- 11:00-11:30 - Coffee break (includes Demos/Posters)
- 11:30-13:00 - Session 2 - Modelling Multimodal Systems
- 4. Agostino Nuzzolo (University of Roma 2 ), “Modelling Intelligent Multimodal Transit Systems” ➕
In recent years advances in information technology and telematics, supporting real-time transit operations control and traveller information, are helping transit agencies achieve more efficient public transportation systems.
The introduction of global positioning system (GPS) mobile devices enables not only real-time individual information to be provided en route, but also traveller location to be received. This means that travellers can be tracked, representing a milestone in the evolution of real-time information.
At the same time, bi-directional communication and big data processing allow advances in real-time short-term forecasting, both of transit network conditions and of traveller numbers on board and at stops. Such data can be used in applications of operations control strategies to improve transit vehicle trip regularity and mitigate crowding. This type of information can also be provided to travellers, who can choose to skip overloaded runs and wait for less crowded ones, thereby making a trade-off between a longer waiting-time and higher on-board comfort. The opportunities given by new telematics applications in supporting service operations control are investigated, and the developments of individual real-time traveller info systems are analyzed.
With the availability of large amounts of data and bi-directional communication, researchers are being encouraged to develop new modelling approaches for real-time transit network simulation, with methods which upgrade and update demand flows and path choice model parameters in real time. Further, traveller-tailored models, which take into account personal preferences, can be defined.
The large quantity of available data can also be used to investigate the scheduling problem of transit services in multimodal transit networks and to design transit networks with stochastic demand.
- 5. Monica Menendez (ETH Zurich ), “Modelling Unimodal and Multimodal Urban Networks: A Macroscopic Perspective” ➕
In recent years, the macroscopic fundamental diagram (MFD) - a well-defined and reproducible relationship between vehicle accumulation and network flow – has gained a lot of popularity, especially for control purposes. Unfortunately, due to the limited availability of empirical data, most of the research on the topic has been limited to simulation environments. Here we will discuss different estimation methods to overcome this limitation. Additionally, we will combine data from loop detectors and automatic vehicle location devices (AVL) of the public transport vehicles to present the first empirical multimodal MFD (i.e., 3D-MFD). We will also show how with limited data, we can still quantify the macroscopic interactions between private and public transport using a statistical model. Last but not least, we will illustrate how to compute the optimal share of public transport users, and how this could be potentially linked to different topological features of the network.
- 6. Jean-Patrick Lebacque (IFSTTAR ), “Multimodal modelling for cooperative ITS” ➕
Analysis and management on a large scale of complex cooperative transportation systems with possibly autonomous agents requires new modelling tools. The object of the presentation is to show that it is possible to build such models in an efficient way, based on the GSOM (generic second order modelling approach). These models were first developped for vehicular traffic modelling. Thus they retain the capacity to represent the dynamics of vehicles at a large scale and to integrate various operational constraints and features, such as infrastructure capacity, speeds, congestion, etc. Public transportation (trains, buses, trams), as well as various types of vehicles (autonomous, electrical ...) can easily be accomodated into this framework. But these models also integrate passenger attributes such as passenger load, allowing for a dual vehicle/passenger representation of flows. Finally GSOM models can also keep track of the information flow, which completes the representation of the cooperative transportation system.
- 13:00-14:30 - Lunch
- 14:30-16:00 - Session 3 - Smart and Sustainable Mobility Systems
- 7. Gaetano Fusco (University of Roma 1 ), “Sustainable Smart Urban Mobility Systems Design” ➕
- 8. Wolfgang Niebel (DLR ) “Challenges in static data acquisition for a bus-on-demand line” ➕
- 9. Marialisa Nigro (University of Roma 3 ), “E-GO: An Innovative Electric Car Sharing Project in Roma Tre University” ➕
- 16:00-16:30 - Coffee break (includes Demos/Posters)
- 16:30-18:30 - Session 4 - Mobility Patterns Extracted from Big Data
- 10. Emmanouil Chaniotakis and Constantinos Antoniou (Technical University of Munich ), “Inferring Activities from Social Media Data” ➕
Social Media have been found to produce an unprecedented amount of information that can be extracted and used in transportation research with one of the most promising areas to be the inference of individuals’ activities. While most studies in the literature focus on the direct use of Social Media data, this study presents an efficient framework for the inference of users’ activities from Social Media data following a user–centric approach. The framework is applied to data from Twitter, combined with inferred data from Foursquare that contains information about the type of location visited. Then, the users’ data is classified using a density–based spatial classification algorithm that allows for the definition of commonly visited locations and the individual–based data is augmented with the known activity definition from Foursquare. Based on the known activities and the Twitter text, a set of classification algorithms is applied for the inference of activities of tweets. The results are discussed upon the types of activities recognized and its classification performance. The classification results allow for a wide application of the framework in the exploration of the individuals’ activity space.
- 11. Chris Tampere (KU Leuven ), “Identifying the aggregate supply for ridesharing through the mobility patterns observed in big data collected by smartphones” ➕
Context-aware applications assist mobile users autonomously when a new requirement is expected in the short-term. Context consists not only of location-based information but of the big data collected by tracking applications from numerous heterogeneous sources including a device sensors. Knowledge about a user’s mobility is acquired by mining over multiple dimensions a travel history at regular intervals, then relevant patterns can be learned including but not limited to frequent destinations, travel modes and arrival times. Ridesharing can benefit from this information, when groups of users are expected to behave similarly. Short-term trip prediction through the combination of assimilated knowledge and information about a user’s recent activity, allows the aggregate supply to be discovered, increasing the opportunities to match a passenger’s future request. The underlying data mining process, strengthens previous discovered patterns and weakens those barely observed in the recent history, iteratively improving the mobility activity detection.
- 12. Raphael Frank (Motion-S ), “The Power of Mobility Data” ➕
In this talk I am first going to describe a driving campaign conducted in Luxembourg in partnership with a local insurance company. During this campaign, a large amount of GPS data has been collected by more than 5.000 users using a Smartphone app. Next, I will present how this raw GPS data can be augmented to create so called “Driving Profiles”. Those profiles not only describe the driving dynamics (acceleration/braking/steering) of a user but also include contextuel information such as driving environment, weather and much more. I will conclude the talk by describing potential applications and discuss privacy implications.
- 13. Thierry Derrmann (University of Luxembourg ), “Estimating Macroscopic Fundamental Diagrams from Mobile Network Data” ➕
- 18:30-19:30 - Cocktail
- 19:30-22:30 - Dinner
- 08:30-09:00 - Registration, welcome coffee & breakfast
- 09:00-10:30 - Session 5 - Autonomous & Connected Vehicles
- 14. Jerome Harri (EURECOM ), “Coexistence Challenges between Radio Local Access Networks (RLANs) and ETSI ITS-G5 in the 5.9GHz ITS band for Future Connected Vehicles” ➕
The increasing need for future WiFi gigabit (IEEE 802.11ac) and LTE Licensed Assisted Access (LTE-LAA) links require to use any available ISM spectrum. The ITS 70Mhz spectrum at 5.9GHz currently reserved for vehicular traffic safety-related applications using the ETSI ITS-G5 technology is therefore under active discussion to be open to other Radio LAN technologies. In this talk, we first present the different technologies that are expected to coexist, describe their inherent differences and formulate the coexistence challenge in a basic urban scenario. Second, focusing on the WiFi technology, we present and evaluate two coexistence protocols proposed by the WiFi community, illustrate the WiFi potential ‘harmful’ interferences to safety-related ITS applications, and propose improvements for a better coexistence. We finally put this work in perspective of current standardization efforts in the ETSI and the design of future RLAN-based connected vehicles.
- 15. Claudio Roncoli (Aalto University ), “Multilane traffic control in the presence of automated and connected vehicles” ➕
A widespread appearance of connected and automated vehicles is expected in the near future. This opens new possibilities for traffic management, via opportune exploitation of the novel possibilities of using vehicles both as moving sensors (providing information on traffic conditions) and as actuators. However, in order to achieve improvements in traffic flow efficiency, appropriate studies, developing potential control strategies to exploit the potential of connected and automated vehicles, are essential.
This talk presents a framework for the coordinated and integrated control of a traffic system, considering that an amount of vehicles are connected and automated. The concept employs and exploits the synergistic (integrated) action of a number of old and new control measures, including ramp metering, vehicle speed control, and lane changing control at a macroscopic level. The effectiveness and the computational feasibility of the proposed approach are demonstrated via macroscopic and microscopic simulations, under realistic traffic conditions and assumptions, for a variety of penetration rates of connected and automated vehicles.
- 16. Anastasia Spiliopoulou and Markos Papageorgiou (Technical University of Crete ), “Real-time ACC Exploitation for Improved Motorway Traffic Flow” ➕
This study presents an ACC(Adaptive Cruise Control)-based traffic control strategy which aims to adapt in real time the driving behavior of ACC-equipped vehicles to the prevailing traffic conditions so that the motorway traffic flow efficiency is improved. The potential benefits obtained by applying the proposed control concept are demonstrated for different ACC penetration rates by use of validated microscopic simulation applied to a real motorway stretch where recurrent traffic congestion is created due to an on-ramp bottleneck. The simulation results demonstrate that, even for low penetration rates of ACC vehicles, the proposed control concept improves the average vehicle delay and fuel consumption by reducing the space-time extent of congestion compared to the case of only manually driven or regular ACC vehicles.
- 10:30-11:00 - Coffee break (includes Demos/Posters)
- 11:00-12:30 - Session 6 - Assessing Future Smart Mobility Services
- 17. Francesco Ciari (ETH Zurich ), “Simulating shared automated vehicle fleets: what can be done with MATSim, current limitations, and modeling challenges.” ➕
The expected diffusion in the near future of fleets of automated vehicles has already generated a fair amount of research in the field of transportation simulations. This talk reports on the current modeling capabilities and limitations of MATSim, based on examples drawn from past and ongoing research. Furthermore, some reflections on the modeling implications of a transportation system largely based on shared automated vehicles are provided. If vehicles become part of the transportation infrastructure, instead of being mere users of it, the conception of new models might be necessary. Some challenges will be delineated and some possible lines of research to address them suggested.
- 18. Lara Codeca (University of Luxembourg ), “Luxembourg SUMO Traffic (LuST) Scenario” ➕
Due to the lack of realistic, properly-working, and freely-available scenarios, during the PhD I created the Luxembourg SUMO Traffic (LuST) Scenario, a mobility scenario built for the vehicular networking research community. The scenario is based on information from a real mid-size city, with a typical European road topology and mobility patterns. The traffic demand is based on real information provided by various data sources. The level of realism provided by this general-purpose traffic scenario has been evaluated and validated using empirical Floating Car Data, showing that the speed distributions from the mobility traces in the simulations reflect a realistic behaviour. The LuST Scenario is already being used by the research community and plans for future improvement and additional features are primarily driven by the needs expressed by the VANET community itself; it is freely available to the whole research community under an MIT license and is hosted on GitHub (https://github.com/lcodeca/LuSTScenario).
- 19. Roberta Di Pace (University of Salerno ), “Traditional Random Utility models vs Hybrid choice models for assessing environmental impacts of a new technology: the HySolaKit case study” ➕
The aim of the research is to investigate if behavioral models more accurate and effective in simulating new automotive technology adoption, may significantly affect market forecasts and environmental impacts estimation. Since the time and the cost to develop each modelling solution are significantly different and require different types of surveys, the main goal of the paper is to give some insights on the costs and on the benefits of each solution. Moreover, different models are compared in terms of sensitivity to different market scenarios and in terms of estimated environmental impacts on a real case study (the city of Salerno – southern Italy).
- 12:30-13:30 - Lunch
- 13:30-15:30 - Session 7 - Demand Estimation using Big Data
- 20. Jaume Barcelo (Universitat Politècnica de Catalunya ), “A Computational Framework for the Estimation of Dynamic OD Trip Matrices” ➕
Formulations of static traffic or transit assignment models, as well as dynamic models involved in ATIS (Advanced Transport Information Systems), usually assume that a reliable estimate of an OD matrix is available, and constitutes an essential input for describing the demand to estimate network traffic states and short term predict their evolution.
The estimation of time-dependent OD matrices has been usually based on space-state formulations using Kalman Filtering approaches as the most suitable to model dynamic phenomena. Variants of this approach have been explored, using variants of Extended Kalman Filters to deal with the time dependencies of model parameters, which are usually included as state variables in the model formulation.
However, when real-time measurements from Information and Communications Technologies (ICT) are available, e.g. those supplied by Bluetooth/Wi-Fi devices, hypothesis on non-linear traffic flow propagation to estimate travel-times between pairs of points in the network are no longer necessary, since they can be measured by these technologies. If the quality of the time-dependent OD estimates strongly depends on the controllable design factors and, if given a purposely designed detection layout and an associated traffic data collection procedure, the determinant factor is the quality of the input OD seed.
In this talk we discuss in detail the above integrated computational framework, its architecture and main components. We also address the question on whether the data structures and equations of the Kalman Filter, which depend on the network topology and the sensor layout, can be automatically adjusted to situations in which sensors fail or some of the ideally planned sensors are not available at the desired locations. We present and discuss some preliminary computational results for simple networks.
- 21. Ernesto Cipriani and Marialisa Nigro (University of Roma 3 ), “New Perspectives for Dynamic Traffic Demand Estimation and Prediction Adopting Big Data” ➕
Nowadays modern cities are facing social and economic changes that pose several mobility concerns related to limited supply of transportation systems. Such issues cannot be tackled only with planning strategies based on capacity increasing options but have to be integrated, if not replaced, by solutions deriving from the adoption of transport management approaches.
In both cases, accurate estimates on travel demand patterns represent the key information for feeding the simulation process required in the evaluation phase.
Demand information can be derived from advanced surveillance systems providing updated measurements of several heterogeneous data, both in fixed locations and over specific corridors or paths. Modern technologies for transport data collection can be classified according to their spatial coverage in point, point-to-point, and area wide. Each of them corresponds to different types, level of accuracy and robustness of detected data.
Measurements of trajectory samples provided by floating cars moving along a whole network combined with OD samples collected on sub-networks by Bluetooth sensors, or unmanned helicopter (drone), represent an unprecedented amount and typology of data availability that is pushing towards a new research area merging analytical and data driven modeling. This is especially true for short term predictions where the capability of detecting and dealing with non-recurrent conditions occurring on the network is becoming increasingly urgent because of information dissemination guaranteed, for instance, by recent numerous path search tools easily accessible by users on desktops and smartphones via the net.
The contribution looks at the opportunities generated by all these new sources of data within the dynamic demand estimation and prediction.
At first, heterogeneity, accounting for different sets of data providing a wide spatial coverage, will be investigated for the benefit of off-line demand estimation. In an attempt to mimic the current urban networks monitoring, examples of complex real case applications will be reported.
Subsequently, on-line detection of non-recurrent conditions will be recorded, adopting a sequential approach based on an extension of the Kalman Filter theory for short-term demand predictions.
Both the off-line and the on-line investigations will adopt a simulation approach capable of capturing the highly nonlinear dependence between the travel demand and the traffic measurements. Consequently, the possibility of using collected traffic information is enhanced, thus overcoming most of the limitations of current approaches for dynamic demand estimation and prediction in the literature.
- 22. Guido Cantelmo (University of Luxembourg ), “Practical methods for Dynamic Demand Estimation in congested Networks” ➕
- 23. Martin Fellendorf (TU Graz ), “Big data to improve the understanding of mobility” ➕
- 15:30-16:00 - Coffee break (includes Demos/Posters)
- 16:00 - 17:30 - Session 8 - Human Mobility Profiling
- 24. Martin Kracheel and Patrick van Egmond (LuxMobility ), “POSITIVE DRIVE, a gamified mobility tool” ➕
Luxembourg attracts many commuters from the neighbouring countries Belgium, Germany and France, where 42 % of the workforce in Luxembourg is commuting from. Luxemburg City ranks 134th out of 1,064 cities in the world and 75th out of 627 cities in Europe that were taken into account for the INRIX study[i]. People spend in average 33.1hours in congestion per year.
It is before this background that a group of major companies in a business area in Luxembourg found together, under the leadership of IMS and with the help of LuxMobility, in order to tackle local mobility problems.
In order to uncover synergies between the major employers of the district and to develop improved mobility concepts for the area, we decided to use the tracking and tracing application Positive Drive, which was developed within the European Research project TRACE.
In this talk, we present the Positive Drive application and how a mobility tracing and tracking campaign can be set up in a professional, cross-border commuting setting such as the Cloche d'Or in Luxembourg. The application provides a gamification layer, which allows us to provide incentives to users. Rewards in Positive Drive are based on a geographical location, called gamezone. Within the gamezone a campaign consists of multiple elements and involves local stakeholders and sponsors.
Using Positive Drive we collect dense mobility data of a local area that is heavily impacted by cross-border commuters.
- 25. Hans van Lint (TU Delft ), “Urban Mobility Lab - making sense of urban mobility through clustering and visualisation” ➕
In this talk I discuss how clustering and visualisation of data for both car and public transport help reveal overall urban mobility patterns.
- 26. Jun Pang (University of Luxembourg ), “Constructing and Comparing User Mobility Profiles” ➕
We propose a new approach to construct users’ mobility profiles and calculate the mobility similarities between users. We model mobility profiles as traces of places that users frequently visit and use frequent sequential pattern mining technologies to extract them. To compare users’ mobility profiles, we first discuss the weakness of a similarity measurement in the literature and then propose our new measurement. We evaluate our work using a real-life dataset published by Microsoft Research Asia and the experimental results show that our approach outperforms the existing works on different aspects. With the implemented software, MinUS, we provide a platform to manage movement datasets, and construct and compare users’ mobility profiles.
- 17:30-18:15 - Panel discussion - “What is smart mobility?”
- 18:15-19:00 - Closing remarks and final farewell cocktail