Estimation of Human Anxiety Based on Skin Conductance Response and its Application to eHMI Performance Evaluation

Pedestrians often feel anxious because they cannot understand the intentions of autonomous vehicles (AVs). To reduce this anxiety, various external human-machine interfaces (eHMIs) using sounds and lights have been proposed to convey AV intentions. Since there is no established method for evaluating which eHMIs can effectively build trust, it is necessary to measure the effectiveness of these eHMIs. This study establishes a method to estimate human anxiety using skin conductance response (SCR) and machine learning, investigating its potential application for eHMI evaluation. While SCR has been used to estimate anxiety, it has typically targeted severe anxiety requiring treatment, such as panic disorders. This study measures and utilizes SCR under conditions inducing relatively mild anxiety through a virtual reality (VR) high-altitude experience.