Real-time emotional state estimation system for Canines based on heart rate variability

Abstract

Search and rescue (SAR) dogs are widely used to locate victims at disaster sites. The efficiency of SAR missions can be greatly enhanced if a canine's emotional states, including their motivation to search, can be remotely estimated in real time. In this study, we developed a real-time emotion estimation system for canines based on measured electrocardiography signals. This proposed system measures a canine's heartbeat intervals using a specially developed canine suit equipped with an electrocardiography device. Using the online heartbeat interval measurements, the system calculates time domain indices of heart rate variability, which are used to classify the canine's emotional state as positive or negative. To support visualization, the system presents the heart rate and estimated emotional state graphically in real time. The real-time emotion estimation system for canines proposed in this study was evaluated using a series of experiments. Bland-Altman analysis showed that online heartbeat interval measurements were consistent with offline heartbeat interval measurements when the canine was at rest or standing still. The proposed system was confirmed operational outdoors in real time and thus has the potential to enhance the efficiency of SAR missions that use canines.