In this project, we developed a taxonomy of human errors in driving on behalf of the Federal Highway Research Institute BASt. Human error can, among other things, lead to accidents. We defined accidents according to a systemic approach, namely as a consequence of the interaction of system components and processes, or applied to the vehicle context, of driver, driving task, vehicle and environment. The classification was not only based on accidents as a consequence of errors, but also on behaviors that almost led to accidents as well as normal, accident-free driving behavior.

In the project, we combined the review of the current state of research and theory building on error classification in vehicle driving with our own empirical investigations. To test the model assumptions of the created error taxonomy, a re-analysis of data from studies on Naturalistic Driving (e.g. 100CarStudy, PROLOGUE) as well as a validation study in a simulator were conducted. Naturalistic Driving Studies investigate normal driving under real-world conditions. They thus offer an extension of the usual analysis basis (accident reports) to include accident-free driving behavior and situations with events that have just not yet led to an accident. Finally, a simulator study was used to investigate partial aspects of the taxonomy in a systematic and controlled manner. The project results supported the selection of measures for error prevention, in particular the derivation of the need for driver assistance.