In many cases, the drivers themselves are responsible for accidents. What kind of support do they need then to reduce this risk? And how can Advanced Driver Assistant Systems ideally be integrated into the vehicle? Studies within the natural driving environment can vitally contribute to answering such questions.

Naturalistic Driving Studies originated in the USA, due to an increase in technological equipment of vehicles - especially with Advanced Driver Assistant Systems. In order to prospectively apply such studies in Germany as well, I-FAS in cooperation with the Research Association for Automotive Technology (FAT) is intensively doing research on further requirements for realization and usage of Naturalistic Driving Studies.

Typical approach to Naturalistic Driving Studies

Several test vehicles are equipped with extensive measurement technologies (e.g. cameras, sensors, eye and head tracker, GPS, CAN-Bus, accelerometer) and handed over to a number of participants. The probands are informed about the procedure of the study and are told what sort of data will be collected. Beyond that, they are not briefed, but expected to use the vehicles as naturalistic as possible over a longer period of time in their every-day life.
 

Image 2: Infrastructure for collecting and transmitting data in Naturalistic Driving Studies

In the course of this ordinary usage of the vehicles, a number of different data is collected. These data and log files are transmitted to a server while the vehicle is standing at its basis (see image 2). A data base collects all incoming data, which can subsequently be evaluated by means of a special software (see image 3).

Image 3: Using a special evaluation software, the information is evaluted by interdisciplinary teams (e.g. psychologists and engineers).

The potential of Naturalistic Driving Studies

General aim of these studies is to detect how people use vehicles in their every-day life and where they are in need of assistance. Further, they seek to gain information about the impact of technological systems on the driving behavior. This allows to draw conclusions on how Advanced Driver Assistant Systems can effectively be applied.

There already are some interesting results available. In the course of a large-scale Naturalistic Driving Study at the Virginia Tech Transportation Institute it was for instance found out that driver drowsiness increases the risk of crashes or other safety-related incidents by six to eight times. Similarly, inattention augments the risk of accident. While secondary tasks like eating, drinking or talking were found to be comparatively unproblematic, looking away from the street or dialing on the phone extensively increase the risk of accident. Furthermore, teenagers were found to be periled by four times compared to adult drivers. Studies at the joint research center "Safer" in Sweden on the other hand have brought important insights concerning data collection and evaluation.

By means of our project, we seek to provide methodical and technical guidelines for future research.
This includes for instance the question, how the effort concerning technology and planning can be reduced. The depiction and discussion of technical approaches applied is another important part of our project. Furthermore, we want to deduce recommendations concerning the requirements on samples: How many driving samples are necessary? How should they be composed? Which recommendations can be given concerning the vehicles used and the observation period? With regard to the data material to be recorded, to data reduction or to statistical evaluation, our poject also seeks to provide important findings in order to improve future naturalistic driving studies. By means of available data we discuss how the significance (validity and reliability) of hardly standardized data on driving behavior can be augmented and which methods are most adequate for evaluation. Drawing upon reanalysis of existing data material, we can further discuss contraditory conclusions and identify possible differences between laboratory experiments and field studies. Finally, our project is supposed to define areas requiring further research.