SWOV Catalogus

332584

Recommendations for a large-scale European naturalistic driving observation study. PROmoting real Life Observations for Gaining Understanding of road user behaviour in Europe PROLOGUE, Deliverable D4.1.
20111309 ST [electronic version only]
Sagberg, F. Eenink, R. Hoedemaeker, M. Lotan, T. Nes, N. van Smokers, R. Welsh, R. & Winkelbauer, M.
Oslo, Institute of Transport Economics TØI, 2011, 105 p., 38 ref.; EU Seventh Framework Programme; Theme 7 Transport / Grant Agreement Number: 233597

Samenvatting Naturalistic driving observation is a relatively new research method using advanced technology for in-vehicle unobtrusive recording of driver (or rider) behaviour during ordinary driving in traffic. This method yields unprecedented knowledge primarily related to road safety, but also to environmentally friendly driving/riding and to traffic management. Distraction, inattention and sleepiness are examples of important safety-related topics where naturalistic driving is expected to provide great added value compared to traditional research methods. In order to exploit the full benefits of the naturalistic driving approach it is recommended to carry out a large-scale European naturalistic driving study. The EU project PROLOGUE has investigated the feasibility and value of carrying out such a study, and the present deliverable summarises recommendations based on the PROLOGUE project. A matrix of research topics and questions has been developed, based on categories of driver behaviour and states combined with various situational categories. The design of the large-scale study should be based on a selection of research questions judged to be particulary important for policymakers and other stakeholders. At the same time it should be acknowledged that the enormous amounts of data that can be collected in a naturalistic observation study can be used both for analysing predefined research questions and for post hoc investigations of new research questions. In designing a large-scale study an optimal trade-off has to be sought between the sophistication of recording technology on the one hand, and the size of the driver sample and duration of the study on the other, since both highly sophisticated technology and a large database imply high project costs. The levels of recording technology may vary from basic driving parameters like speed, acceleration and position to a system including several video channels, eye-tracker and a multitude of vehicle parameters. An important purpose of naturalistic driving observations is to identify crash-related behaviour and to estimate the relative risks associated with those behaviours; in other words, to validate various crash proxies or crash surrogate measures to be used e.g. in subsequent monitoring of safety performance indicators. Since crashes are rare, this requires large samples and/or long duration of the study, amounting to several millions of vehicle kilometres. It has to be carefully considered which parameters are the most important to observe in such a large-scale study to capture the potentially most relevant crash-related behaviours. One possibility is to design a multi-level study, consisting of a very large sample with recording of basic driving parameters, a medium-size sample with some more advanced recording technology (e.g. trigger-based video to record pre-defined types of incidents) and a small sample with very advanced recording technology (including e.g. continuous video and eye-tracking). When recruiting participants to the study, representativity as well as possibilities of investigating behaviours of particular groups of drivers must be considered. Other issues that need attention are personal privacy protection of both drivers and passengers, and participants’ right to withdraw from the study. Monetary incentives as well as adequate procedures to avoid inconvenience to participants may be important measures for successful recruitment. There are several important requirements to the data acquisition system, including reliability, ease of installation and deinstallation, and the possibiliity of transferring data effectively, without data loss, and without inconvenience for the participants. Finally, adequate tools for data analysis are required, including good algorithms and procedures for identifying critical incidents and driver actions to be analysed. (Author/publisher) See http://www.prologue-eu.eu/ for more information and other reports about this programme.
Full-text
Dossier
Suggestie? Neem contact op met de SWOV bibliotheek voor uw opmerkingen
Copyright © SWOV | Juridisch voorbehoud | Contact