A driver may appear attentive while their ability to respond safely is already impaired. Current driver monitoring systems detect visible signs of distraction, drowsiness, and unresponsiveness, but broader physiological or cognitive impairment may not be immediately observable through behavior alone. Severe fatigue and alcohol-related impairment are important target scenarios for investigating this safety gap. This presentation introduces our exploration of a vision-based, multi-domain framework combining behavioral observations, camera-derived physiological information, and driving-response evidence. Our initial focus is to assess whether consistent evidence indicates a meaningful reduction in driving readiness, while also investigating whether different impairment conditions can eventually be distinguished. We will discuss the potential of the existing near-infrared driver monitoring camera, the challenges of obtaining reliable evidence in real cabin environments, and a staged validation approach for evaluating generalized impairment before progressing toward condition-specific assessment.
