The objective of the Task 6.2 of DaCoTA is to specify the study design of naturalistic driving study in the perspective of the European Road Safety Observatory. More precisely, the task deals with three main issues: 1) the experimental design, 2) the procedures to Risk Exposure Data (RED) and Safety Performance Indicators (SPI) estimation, 3) legal, ethical and privacy requirements. The Deliverable 6.2.A "Study design of Naturalistic Driving observations within ERSO - Development of innovative indicators for exposure and safety performance measures" aims to provide operational guidelines and recommendations that can be used to take maximum advantage of naturalistic driving data to observe road safety related phenomenon. General design constraints are discussed. The design of new RED and SPI using Naturalistic Driving Data (NDD) has to be done by bringing added value to existing RED/SPI that can be obtained by classical investigation means (observation from the side of the road, surveys...). The design must also take into account the link between the data collection solution and the data available for the indicator computation as this has an important impact on the possible RED and SPI that can be measured. Key principles about the possibilities of data aggregation (to present results at different levels), data filtering (to obtain homogenous driving conditions) and data clustering (to compare different conditions) are presented. The question of the accuracy of the results is also addressed. Some recommendations and best practices are given in order to establish some reference definitions of the key elements that may impact the RED/SPI results: obtaining homogeneous driving data in term of driving situation, aggregation using "trips", comparisons according to "night" and "day" conditions or "good" and "bad" weather conditions, systematic calculation of the "sub sample characteristics" to assess the amount of data used for the indicator computation... Concerning the RED, authors propose a list of innovative RED and point out the limitations of the NDD that should be taken into account when interpreting the RED results. An accurate description of the process to compute the RED is proposed. Concerning the SPI, authors propose a distinction of 3 kind of SPI: behavioural SPI, descriptive SPI and situational SPI. They also propose a list of innovative SPI and point out the limitations of the NDD that should be taken into account when interpreting the SPI results. An accurate description of the process to compute the SPI is proposed. The aim of the RED/SPI description chapters is to gather all required information so that this part can be used as a practical handbook for SPI/RED development in the frame of ERSO. However, the aim of the RED/SPI description chapters was not to give the description of all possible calculations that could be made using NDD. Authors focused on a set of RED/SPI that they considered as the most relevant. Using the best practices and recommendation, it makes it possible to compute new SPI and RED that have not been described in detail. The question of the driver sample of driver to monitor is also addressed. This aspect is discussed more in detail in deliverable 6.2.B which proposes an extensive discussion of the sampling methods and impact on the results. Some recommendations are given to help with the relevant criteria for the recruitment of the participants. D6.2 Study design of Naturalistic Driving observations within ERSO - Development of innovative indicators for exposure and safety performance measures Finally, the key information concerning the legal and ethical issues is given, as it has to be taken into account, both at global and local level, when implementing the NDS. (Author/publisher)
Abstract