Analysis of driving behaviors based on GMM by using driving simulator with navigation plugin

Generally, novice drivers often feel uneasy about driving in unfamiliar traffic environment. In Japan, the road traffic act was revised in November, 2011. This revision authorized the operation of the roundabout intersection which is a traffic-lights-free intersection. The roundabout intersection have been widely spread in Europe, on the other hand, Japanese people are unfamiliar with the roundabout intersection, and especially novice drivers may panic easily. Thus, this paper aims for modeling driving behaviors to support operation of novice drivers. Moreover, we examine the effects of navigation at the roundabout intersection for the novice drivers. First, we built a driving simulation environment for roundabout intersection by using UC-win/Road, and we collect driving logs of a novice driver who drives with navigation or without navigation. Then, we classify the driving log into some discrete driving states (i.e., go straights) on the basis of Gaussian Mixture Model (GMM) and plot state transition graphs of the discrete driving states to model driver´s behavior. The driver model obtained by above method showed the characteristic driving behaviors of novice drivers in the simulation environment.