Data combination using belief theory to assess driver?s vigilance

Human error has been implicated as a causative factor in 85% of drivers? and operators? crashes, and lack of vigilance has been identified as the single most important factor in incidents involving human error. Driver vigilance could decline with sleepiness, fatigue or monotony. In Queensland, inattention and fatigue respectively contribute to 27% and 5% of reported crashes. Vigilance decline is characterised by an increased or absence of response to critical events. The current technology to assess and prevent vigilance decline is based on the isolate use of a particular device such as eye tracker or steering wheel movements. The reliability of these devices is debatable as the value of the readings could be highly inaccurate, uncertain, partial, conflictual or unreliable. Furthermore, there has been very little research examining the use of multiple devices to diagnose vigilance decline.

The aim of this paper is to use belief theory to assess driver?s vigilance. Belief theory is a formal tool suitable for representing the inaccuracy, uncertainty and asynchnocity of knowledge. Our approach consists of merging a set of measurements, related to the environment, driver, and vehicle, gathered from different Advanced Driver Assistance Systems (ADAS). This paper presents the theoretical basis leading to the development of an advanced in-vehicle system capable of assessing vigilance decline.  The development of such a tool has a potential to be a major contributor to reducing death and injury rates due hypovigilance related driver?s errors.