Automatic Recognition Model of Motion Sickness and Hierarchical Classification by Random Forest of Line of Sight

Abstract

There are many studies on improving motion sickness in cars. Many of the factors focused on a vestibular organs and somatosensory. There are few examples of studying visual information. There-fore, we investigated only movements of a human viewpoint, not the psychological factors. Next, the line-of-sight trend data was automatically analyzed into a directory structure using the random forest. As a result, we clarified the characteristics of people who are prone to motion sickness and those who are less likely to have motion sickness. We created a model based on this feature and confirmed that it is possible to automatically determine people who are prone to motion sickness. This makes it possible to give advice on gaze trends to eliminate motion sickness to people who are prone to motion sickness.