Abstract
Vehicle manufacturers consider human–machine interface customization as an innovative method to increase the usability and application of cockpits and on-board navigation devices. For customization to be applicable, it needs to reflect relevant driver attributes such as preferences and characteristics. In this study, we propose a method to assess personality traits using vehicle motion and driving activity signals as input. First, a pilot survey was conducted to investigate the relationship between driving activity and psychological scale. In particular, comparing the frequency of empathetic activity of drivers with an emotional empathy score, we demonstrated that there was a significant relationship between both values (p < 0.05). Second, we developed a hidden Markov model based statistical classifier. The system comprises two models: the first is used for classifying road configurations and the second is applied to classify sociability and other social interaction related psychological attributes. Based on driving signal inputs, the system calculates probability scores, and it outputs the probable personality trait. To validate the estimation algorithm of activities, we computed road configurations and driver psychological attributes on a reference data set collected during actual driving sessions. The calculation demonstrated 85% accuracy for the road configuration classification. The detection rate for driving activities, such as the empathetic activity, was approximately 70% or greater, and the concordance rate for personality traits reached over 85%.
KEYWORDS – HMI automatic customization, driving activity estimation, emotional empathy scale, driver’s preference