Improved Reliability of Driver Drowsiness Ground Truths with Objective Assessment

The Karolinska Sleepiness Scale - KSS is often used as a ground truth of drowsiness in real-world studies, since it is easy to use, has high acceptance, and is relatively robust to intra-individual differences. Its subjectivity can, however, be a concern. This paper therefore proposes a multimodal framework that integrates subjective self-reports (KSS) with objective cognitive assessments (in this case, the Detection Response Task - DRT) using regression modeling, statistical analysis, and machine learning. Simulator studies show strong correlations for some participants, highlighting the potential of the proposed approach to enhance road safety and ADAS performance.