WP6 of DaCoTA, Driver Behaviour Monitoring through Naturalistic Driving, focuses on the usefulness and feasibility of applying the Naturalistic Driving method for monitoring within the framework of ERSO. The aim is to continuously collect comparable information about the road safety level in EU Member States and its development over time. Naturalistic Driving methods are intended to gather data that represent the behaviour of the population of drivers in its basic state. Naturalistic Driving (ND) study can be defined as: ‘A study undertaken to provide insight in driver behaviour during every day trips by recording details of the driver, the vehicle and the surroundings through unobtrusive data gathering equipment and without experimental control’ Typically, in an ND study passenger cars are equipped with small sensors. These devices continuously and inconspicuously register vehicle manoeuvres (like speed, acceleration/deceleration, direction, location), driver behaviour (like eye, head and hand manoeuvres), and external conditions (like road, traffic and weather characteristics). ND for monitoring purposes ND data can be used to establish how often drivers routinely are exposed to or engaged in certain situations/behaviours that are known to increase the risk of a crash. This includes monitoring safety-relevant behaviour (Safety Performance Indicators - SPIs) and mobility (Risk Exposure Data — RED). Benchmarking is an important reason for monitoring road safety and comparing road safety levels and their developments over time. It allows countries to determine their relative position in comparison to other countries, to understand differences and find ways and get motivated to improve their position. Obviously, monitoring road safety also allow countries to evaluate their own road safety policy and road safety targets. ND is considered a promising approach for collecting reliable and comparable information about various RED and SPIs, as well as several relevant context variables. The main advantage of the ND approach for monitoring as compared to the more traditional SPI data collection methods, such as road-side surveys and questionnaires, is that ND ensures continuous, automatic and standardized data collection. This approach substantially increases international comparability and level of detail. Though the current Deliverable is purely focused on road safety and exposure data, the collected data will also be useful for other transport areas, in particular eco-driving and traffic management. Three data collection scenarios Depending on the variable of interest, ND data collection needs different technologies ranging from simple data acquisition systems to more sophisticated systems with several sensors and several videos. By combining the RED and SPIs of interest and the technological requirements for collecting that type of data, we distinguish three scenarios to collect meaningful data within reasonable limits of cost and complexity. It is recommended to start off with Scenario 1: a low-cost simple, off-the-shelf simple data acquisition system (e.g. an OBD GPS tracker or a Smart Phone) and a limited number of additional sensors, basically measuring RED and speed. At a later stage, successively additional sensors could be added to measure more advanced SPIs and network characteristics (Scenario 2). SPIs that would need continuous video recordings do not seem to be feasible for monitoring purposes, because of huge amounts of data and high costs of data transfer and coding. That means that the SPIs like fatigue, inattention and distraction can currently not be monitored by means of ND research. Technical developments may allow reconsideration of this conclusion in due time. It is recommended to equip a limited number of cars also with an event-triggered video in order to monitor numbers of near crashes as yet another relevant SPI (Scenario 3). Study design and organisational issues In principle, the techniques and procedures for ND data collection, transfer, storage and analysis are available and not too complicated. In order to get reliable information, a fairly large sample is needed. The exact size of the sample depends on the variation in behaviour in the population and the required level of precision of the results. Assuming that the sample is drawn in a cleverly stratified way a sample of around 10,000 drivers per country seems to be required for RED such as the annual mobility. This sample size is independent of the size of the country. Only if the sample size is larger than 10% of the population, a correction is applicable. Participation in ND research is per definition on a voluntary basis and past experiences have shown that it requires special attention to find sufficient suitable participants. There are legal and ethical issues involved in ND research, in particular in the areas of privacy and data protection. Exploring Scenario 4 In parallel to the implementation of the previous three scenarios, it is recommended to start exploring the possibility of a Scenario 4 now, i.e. a scenario where relevant data is extracted directly from systems built-in by the vehicle manufacturers. In theory, that way a lot of relevant information is already available with no or few additional costs; in practice, however, the information is not generally accessible nor comparable between car makes. So this is a scenario that cannot be realised overnight. One of the first steps, in consultation with the car manufacturers, is an elaboration of the requirements for this data: what is available, what is needed, what is feasible. A central role for the EC Since harmonisation and international comparability of data are the key reason for this effort, there is a central role for the European Commission in initiating this task and taking the lead from here, most likely within the ERSO framework. A stepwise approach is recommended, including successively: 1. Creating support and finding budget by presenting the case to the relevant road safety bodies at European and Member State level, explaining the need for harmonised, comparable international data, the ND approach, and its added value. 2. Preparing a detailed description of all practical implementation aspects, including the functional specifications of data collection equipment, participant selection, data transfer and storage, definitions of variables, disaggregation levels and analyses. 3. Identifying the relevant national organisations, responsible for national data collection and pre-analyses, and fine-tuning data collection procedures (including legal aspects) and variable definitions in consultation with them. 4. Developing and equipping a database at EU level and defining the required data to be provided and the procedures and time schedule, in consultation with the relevant national organisations. 5. Setting up European-wide communication strategies to guarantee maximum dissemination and use of the collected data. 6. Setting up one year national pilots in at least four Member States. 7. Adapting procedures and definitions, based on the pilot experiences. 8. Successive implementation of Scenario 1 in additional Member States. Parallel to steps 6 and 7, Scenario 2 (additional SPIs/RED) and 3 (monitoring nearcrashes) can be elaborated, piloted and implemented, applying a similar stepwise process. From the very beginning, the EC is advised to initiate discussions with the car manufacturers, using existing discussion platforms, with the aim to explore longer term possibilities of Scenario 4, i.e. the scenario where relevant data is extracted directly from the vehicle. (Author/publisher)
Abstract