"FaceLight": potentials and drawbacks of thermal imaging to infer driver stress

Driving a modern vehicle is a complex, cognitive demanding task involving concentrated observation of the road, roadside, car status, information displays of assistance systems, etc. Drivers are conscious about this overabundance of information, nevertheless, they are operating tertiary controls, talking on the phone, smoking cigarettes, having lunch, reading maps or meeting agendas, or working on their computer. As a consequence -- caused by visual/manual/cognitive demand and limited multitasking capabilities-- precarious driving situations are created. Solutions are rare but badly needed to prevent imminent danger on the roads. To explore the potential of thermal imaging to infer mental conditions of the driver in an unobtrusive manner, and to use this information to automatically react to a detected risky state, we have developed the "FaceLight" prototype and performed a lab-based driving simulator study to evaluate the interface under conditions of varying workload. With "FaceLight" the driver can be interpreted as sort of signal light, with a 'red face' (hot surface temperature) standing for high stress or cognitive demand while a 'green face' (cooler temperature) equals to a more relaxed, stress-free mental state. Initial results have revealed that this technology has potential to capture shifts in the mental state of an individual in an inattentive manner, but highlighted also that a lot of influencing factors still need to be incorporated to reliably recognize a specific state solely based on facial skin temperature.