Structurization with probabilistic causal model for unsafe driving behavior of older people at stop sign intersections

Abstract

To reduce unsafe driving among older drivers, clarifying the process and factors leading to such behaviors is essential. In this study, we proposed a method to express the process and factors leading to unsafe driving using Bayesian networks. In the proposed method, the flow leading to unsafe driving is divided into three stages: driver characteristics, driving behavior, and driving unsafeness. Driver age, various driving behaviors, traffic violation and collision risk were selected as indices of driver characteristics, driving behavior, and driving unsafeness. By organizing and layering these indices in the order of cause, behavior and result, the indices were set to align with the timeline. With the proposed structuring method, we constructed models of deceleration and intersection-passing behaviors based on the driving data of older drivers at stop-sign intersections. The results revealed that driving characteristics and environmental factors influence indices related to the position and speed at intersection entry, capturing relationships leading to violations at stop lines and collision risk with crossing vehicles at intersections. Moreover, it was confirmed through probability inference that the association shown in the structured results was identical to the known driving characteristics of older drivers. According to this result of probability inference, the validity of the proposed structuring method was confirmed.

Keywords

• driving behavior
• older people
• bayesian networks(bn)
• unsafe driving
• driving evaluation