A dynamic avoidance mobility model for the following car using naturalistic driving data

Abstract

With the improvement of automated vehicle technology, it is expected that automated vehicles and Human-driven vehicles will share road traffic for a long time in the future. For some low-speed autonomous driving vehicles (<14km/h), which developed for the elderly, one obvious problem is that if they are driven on public roads for a long time, they will affect the normal traffic flow, reduce the efficiency of traffic, and also cause a decrease in social acceptability. In this paper, we aim to study of avoidance strategy for low-speed automated vehicles to give the way for the approaching rear vehicles, which is expected to avoid the dissatisfaction of the following vehicle as much as possible, and without a large delay in the arrival time of the ego-vehicle. A vehicle deceleration behavior dataset is developed by using the naturalistic trajectory data of human drivers collected on the Urban Streets in the United States, and extract the deceleration behavior of normal drivers when they noticed a slower car ahead and built those data into a new dataset. Based on the regression analysis of the database, we propose a novel dynamic avoidance mobility model for the following car. It can dynamically adjust the timing of avoidance according to the velocity and headway of the following car. The experimental results show that our method can achieve a relatively low delay of the ego-car arrival time while the proportion of the rear driver’s dissatisfaction is kept at a low level