Towards monitoring and modelling for situation-adaptive driver assist systems

Abstract

In the classic tri-level study of the causes of traffic accidents, (1979) ascribe 92.6% of car accidents to human error, where human errors include improper lookout (known as 'looking but not seeing'), inattention, internal distraction and external distraction. (2003) reports that other studies have found similar results that a human error is involved in 90% of car accidents. Human errors, such as those listed in the above, can happen for everybody and may not be eradicated. However, if there were some technology to detect driver's possibly risky behaviour or state in a real-time manner, car accidents may be reduced effectively. Proactive safety technology that detects driver's non-normative behaviour or state and provides the driver with appropriate support functions plays a key role in automotive safety improvement. Various research projects have been conducted worldwide to develop such technologies (see, e.g. Witt, 2003; Panou et al., 2005; Saad, 2005; Amditis et al., 2005; Tango and Montanari, 2005; Cacciabue and Hollnagel, 2005). © 2007 Springer-Verlag London.