自動車運転中の見落とし防止システムの開発にむけて

抄録

We perceive different visual images even when we are in the same place and looking at the same visual scene. This is mostly due to the fact that retinal image of the identical visual scene can be different as characteristics of eye optics, eye movements, pupil and accommodative fluctuations, and spatial arrangement of retinal photoreceptors are all unique in each individual. Thus a driving scene that is easy to manage for some people may be dangerous for others who are prone to overlook important objects due to degraded quality of their retinal image by myopia, presbyopia, retinal detachment, improper gaze direction, or/and other factors. As such, to prevent car accidents caused by oversights, we have been developing a software system that estimates retinal image of each car driver, then quantifies risks of oversight on each part of visual scene, and thereby enables a personalized assistance. In this talk, I will introduce how the system calculates individual retinal image, and show examples of estimated time-varying retinal images during driving simulation as well as real car driving.