A new methodology to overcome memoryless property of car-following models

Taehyung Kim, David J. Lovell, Yongjin Park, and Myungsoon Chang

Journal of the Eastern Asia Society for Transportation Studies, Vol. 5, pp. 1194-1210.

ABSTRACT

A number of different types of car-following models have been developed and continuously refined up to the present time, with various different approaches of describing a relationship between the leader and the following vehicles. However, current car-following models are “memoryless” in the sense that the current state (which consists of instantaneous relative spacings and time-derivatives thereof) between the lead and the following vehicles entirely determines the future state of the following vehicle, with no dependence on the past sequence of events that produced that state. This study aims to propose a new methodology that considers the past sequences of car motions to predict the following vehicle’s behavior and to overcome the memoryless property of previous models on car-following behavior.