Drowsy Bather Detection Using a Triaxial Accelerometer

Japanese people like soaking in bathtubs; however, the number of deaths by drowning among those over the age of 65 is increasing year by year. We focused on the problem that there are no popular solutions for those accidents caused by nodding off while bathing. Today, lightweight wireless earbuds often feature accelerometers. Further, more and more people enjoy music with such earbuds while bathing. Therefore, if a person’s drowsiness in the bathtub can be detected using these accelerometers, wireless earbuds worn while bathing may prevent serious accidents before they happen. This paper proposed a drowsy bather detection system using an accelerometer. When a person nods off, their head tilts gradually, and the direction of gravity applied to the head changes. The system detects drowsiness by detecting these changes with the accelerometer. In the experiment, the usefulness of the algorithm in the system was evaluated by comparing the drowsiness level estimated by the system with by data from a sleep tracking device. The detection rate tended to increase as the stage of non-REM sleep increased. These results suggest that the system may be useful for preventing accidents due to nodding off while bathing.