Prediction of Driver's Workload from Slow Fluctuations of Pupil Diameter

Abstract

Driver's mental workload affects driving performance. The safety technology systems equipped on vehicles need to infer the mental workload in order to decrease driver's stress and provide safety to drivers. The mental workload could be evaluated by physiological data such as slow fluctuations (LF and HF) of heart rate. LF and HF can be estimated from pupil diameter variations in same frequency band as heart rate. In this study, we measured pupil diameter variations while participants drove on the general road and metropolitan expressway. After the experiment, the participants scored their mental workload levels. Then we tested whether their subjective mental workload levels were classified from slow fluctuations in left and right pupil diameter variations by the linear support vector machine. The classification accuracies were significantly higher than the chance level except for one participant. Our results suggest that driver's mental workload levels may be predicted from slow fluctuations of pupil diameters in daily driving.