Thierry Derrmann

Driver Behavior Profiling Using Smartphones: A Low-Cost Platform for Driver Monitoring

Todays smartphones and mobile devices typically embed advanced motion sensors. Due to their increasing market penetration, there is a potential for the development of distributed sensing platforms. In particular, over the last few years there has been an increasing interest in monitoring vehicles and driving data, aiming to identify risky driving maneuvers and to improve driver efficiency. Such a driver profiling system can be useful in fleet management, insurance premium adjustment, fuel consumption optimization or CO2 emission reduction. In this paper, we analyze how smartphone sensors can be used to identify driving maneuvers and propose SenseFleet, a driver profile platform that is able to detect risky driving events independently from the mobile device and vehicle. A fuzzy system is used to compute a score for the different drivers using real-time context information like route topology or weather conditions. To validate our platform, we present an evaluation study considering multiple drivers along a predefined path. The results show that our platform is able to accurately detect risky driving events and provide a representative score for each individual driver.

Smartphone-Based Adaptive Driving Maneuver Detection: A Large-Scale Evaluation Study

The proliferation of connected mobile devices together with advances in their sensing capacity has enabled a new distributed telematics platform. In particular, smartphones can be used as driving sensors to identify individual driver behavior and risky maneuvers. However, in order to estimate driver behavior with smartphones, the system must deal with different vehicle characteristics. This is the main limitation of existing sensing platforms, which are principally based on fixed thresholds for different sensing parameters. In this paper, we propose an adaptive driving maneuver detection mechanism that iteratively builds a statistical model of the driver, vehicle, and smartphone combination using a multivariate normal model. By means of experimentation over a test track and public roads, we first explore the capacity of different sensor input combinations to detect risky driving maneuvers, and we propose a training mechanism that adapts the profiling model to the vehicle, driver, and road topology. A large-scale evaluation study is conducted, showing that the model for maneuver detection and scoring is able to adapt to different drivers, vehicles, and road conditions.

Towards characterizing Bluetooth discovery in a vehicular context

Bluetooth has, in recent years, gained more and more momentum. New commodity objects and wearables implementing Bluetooth Smart technology (Low Energy) are released everyday. In particular, the ever increasing number of discoverable devices both inside and outside a populated area gives us an encouraging insight on future research directions for this technology. In this paper, based on a sensing system developed as an Android application, we evaluate Bluetooth Classic and Low Energy discovery characteristics from a vehicular perspective. By recording information about devices nearby (e.g. the number of discovered devices, their signal strength, manufacturer information) and the GPS location we can derive interesting information about a drivers situation, as well as his/her environment. Presented results indicate that the amount of discovered devices and signal strengths are dependent on velocity and road category. Finally, future work and discussions address potential use-case applications based only on Bluetooth discovery, such as low energy and privacy friendly road and traffic context awareness. The sensing system used in this article is free online under the MIT License.

Validation study of risky event classification using driving pattern factors

In recent years, due to the increasing sensing capabilities of mobile devices, smartphones have become a suitable solution for telematics systems. By combining multiple sensors, GPS information and environmental data, smartphones can be used to detect abnormal driving events that are used to compute driving scores. In this paper we propose a Multivariate Normal model for abnormal driving events detection. This model takes as input smartphone motion sensors and GPS data and detects abnormal driving maneuvers that are classified as events of three different classes: acceleration, braking and cornering. Based on these events, a driving score is computed. In order to validate the reliability of the computed scores, we propose a correlation analysis of the driving score against multiple well-known driving pattern factors proposed in the literature.

Poster: LuST-LTE: A simulation package for pervasive vehicular connectivity

Recent technological advances in communication technology have provided new ways to understand human mobility. Connected vehicles with their rising market penetration are particularly representative of this trend. They become increasingly interesting, not only as sensors, but also as participants in Intelligent Transportation System (ITS) applications. More specifically, their pervasive connectivity to cellular networks enables them as passive and active sensing units. In this paper, we introduce LuST-LTE, a package of open-source simulation tools that allows the simulation of vehicular traffic along with pervasive LTE connectivity.

How mobile phone handovers reflect urban mobility: A simulation study

We propose a novel way of estimating Macroscopic Fundamental Diagrams (MFD) (or often also called Network Fundamental Diagrams) from mobile phone signaling data under the assumption that vehicles can be identified from the data stream. We run a simulation study to identify whether MFDs can be constructed from this type of data. We co-simulate the road traffic in Luxembourg City and one mobile operators LTE network user plane, and show that mobile network base station clusters cover road network partitions of coherent behavior. Our results indicate that the relationships between handovers (flow) and attached phones (accumulation) in these base station clusters constitute MFDs. We validate our results by comparing the road network MFDs to the ones obtained from the mobile network.

Effectiveness of the two-step dynamic demand estimation model on large networks

In this paper, the authors present a Two-Step approach that sequentially adjusts generation and distribution values of the (dynamic) OD matrix. While the proposed methodology already provided excellent results for updating demand flows on a motorway, the aim of this paper is to validate this conclusion on a real network: Luxembourg City. This network represents the typical middle-sized European city in terms of network dimension. Moreover, Luxembourg City has the typical structure of a metropolitan area, composed of a city centre, ring, and suburb areas. An innovative element of this paper is to use mobile network data to create a time-dependent profile of the generated demand inside and outside the ring. To support the claim that the model is ready for practical implementation, it is interfaced with PTV Visum, one of the most widely adopted software tools for traffic analysis. Results of these experiments provide a solid empirical ground in order to further develop this model and to understand if its assumptions hold for urban scenarios.

Towards privacy-neutral travel time estimation from mobile phone signalling data

Todays mobile penetration rates enable cellular signaling data to be useful in diverse fields such as transportation planning, the social sciences and epidemiology. Of particular interest for these applications are mobile subscriber dwell times. They express how long users stay in the service range of a base station. In this paper, we want to evaluate whether dwell time distributions can serve as predictors for road travel times. To this end, we transform floating car data into synthetic dwell times that we use as weights in a graph-based model. The model predictions are evaluated using the floating car ground truth data. Additionally, we show a potential link between handover density and travel times. We conclude that dwell times are a promising predictor for travel times, and can serve as a valuable input for intelligent transportation systems.

Asymmetry Analysis of Inbound/Outbound Car Traffic Load distribution in Luxembourg

The country of Luxembourg possesses a road network topology that provides direct highway connectivities to its three neighbouring countries. Its economic strength attracts cross-border commuters who represent approximately half of the work force in Luxembourg. As a consequence, significant flows of vehicles coming from France, Germany and Belgium significantly contribute to the daily traffic load distribution of Luxembourg. In this paper we propose to analyse the mobility of drivers in Luxembourg, as well as cross-border traffic. We use a loop detector traffic dataset, based on the road traffic counting architecture of Ponts et Chaussées (PCH) recorded in 2010. We propose to describe the behaviour of drivers in the spatial and time domains for the Luxembourg use case. Our first contribution is the analysis of the daily traffic demand profile of each counter and their classification into inbound, outbound and neutral profiles. The evaluation of these demand profiles will help to better understand and meet the actual demand of commuters. Our second contribution is the detection of inbound and outbound anomalies and the analysis of typical cross-border traffic characteristics in Luxembourg.