Detecting car-following and platoons within traffic using the dynamic time warping algorithm

Abstract

The increase in adaptive cruise control (ACC) penetration rate might increase energy consumption, reduce road capacity, and decrease safety levels. Those externalities can be linked with the string instability they create while platooning. To assess ACC’s future impact, it is necessary to know the percentage of platoons within traffic and their length. However, no method exists to quantify platoons. This article proposes a new approach to detect car-following based on dynamic time warping. This methodology is then extended to platooning. The method is then validated on a trajectory dataset both in free-flow and congested conditions. The framework appears promising as it highlights efficiently if a vehicle is car-following or platooning, and it determines the size of the platoon.