表面発光型文字表示に対する生理指標に基づく疲労推定システムの開発

抄録

Light emitting diodes (LEDs) have been widely used for the luminescent text display in automobiles. The purpose of the current study is to evaluate fatigue reduction characteristics in LEDs luminescent text display. For this purpose, we construct a model to estimate the participant's fatigue from physiological index. In the experiment, participants read Japanese sentences that were represented by backlight emitting for ten minutes. Brain activity were measured for the physiological index, and fatigue evaluation questionnaire was answered. The luminescent text display was implemented with LEDs of six kinds of white color. We constructed the deep learning model that inputs brain activity, and outputs “fatigue”, “no fatigue” or “no change”. The output was calculated from fatigue evaluation questionnaire. The model was able to estimate the fatigue with a validation accuracy of 86.4 %. Thus, the model showed the possibility that estimate fatigue based on the brain activity.