Estimating driver awareness of pedestrians from driving behavior based on a probabilistic model

Preventing pedestrians from motor vehicle accidents is a key challenge in achieving a safe and secure society. One advanced driver assistance system detects pedestrians with an in-vehicle sensor and informs or alerts the driver of their presence. If a pedestrian appears on the road, drivers will notice the pedestrian in most cases and adopt appropriate driving behaviors. Therefore, a system that alerts the driver of the presence of a pedestrian whenever a sensor detects a pedestrian would be extremely distracting and annoying. To avoid this problem, we consider a system in which drivers are only alerted when they are not already aware of the detected pedestrian. In this paper, we describe a new approach for estimating the awareness a driver has of pedestrians appearing on the road ahead. It can be assumed that drivers will adopt different driving behaviors when they are aware of a pedestrian. Based on this understanding, we propose a method that uses driving behavior data to estimate whether a driver is aware of a pedestrian.