Towards a Context-Dependent Multi-Buffer Driver Distraction Detection Algorithm

This paper presents initial work on a context-dependent driver distraction detection algorithm called AttenD2.0, which extends the original AttenD with elements from Minimum Required Attention (MiRA) theory. Central to is time buffer keeps track of how often and for long looks away forward roadway. When depleted when looking back fills up. If runs empty classified as distracted. AttenD2.0 this concept by adding multiple buffers, thus integrating situation dependence visual time-sharing behaviour in transparent manner. Also, increment decrement buffers are now controlled both static requirements (e.g. presence an on-ramp increases need monitor sides mirrors) well dynamic (e.g., reduced speed lowers speedometer). The description generic, but real-time implementation concrete values different parameters showcased driving simulator experiment 16 bus drivers, where was used ensure that drivers attentive before taking control after automated stop docking depot procedure. scalability relative available data sources level vehicle automation demonstrated. Future includes expanding real-world environments automatically situational information vehicles environmental sensing digital maps.