Charging demand for the unserved — an agent-based model approach

Abstract

Collision warnings play a crucial role in preventing crashes based on estimating the critical time to potential accidents. Existing research and applications mainly focus on longitudinal vehicle interaction and headway keeping on highways. However, urban driving with frequent lateral interaction may have different critical time to collision. This study considers both longitudinal and lateral interaction through a two-dimensional spacing measure, driver space, to estimate the average critical time for drivers to respond to a potential collision. With the average spacing between vehicles at different levels of discomfort and in different relative speeds, we estimate the critical time via linear regression. Our experiments on two trajectory datasets find that drivers are more alert to collision dangers on highways compared to urban intersections, and drivers respond to potential collisions more quickly during lateral interaction than longitudinal. These findings emphasise the need of tailored collision warning systems for further improving road safety.