We present a video data reduction protocol that was developed to classify the type and intensity of emotion expressed by a driver involved in a cell phone conversation. Although there has been substantial research on coding emotion from a person’s facial expression, the prescribed methods are significantly detailed and require a hundred hours of training as well as certification. The objective of the current research was to develop a simplified protocol based on the previous research that could be applied to naturalistic driving data by data reductionists in a reasonable amount of time to distill information pertaining to the nature of the driver’s cell phone conversation. We present the basis for the protocol and describe how it was applied. In identifying an emotion from a driver’s facial expression using naturalistic driving data, it will become possible to compute how often drivers engage in emotional cell phone conversation, the risk of a Safety-Critical Event (SCE) associated with emotional cell phone conversation, as well as the relative risk of emotional cell phone conversation compared to neutral cell phone conversation. In conclusion, we discuss other related applications of the protocol, how the protocol could be developed further and potential synergy with prior research. (Author/publisher)
Abstract