01592866

Component

Lane departure crashes on horizontal curves accounted for approximately 28% of all fatal crashes in 2008. Curves have been found to have a three times higher crash rate than tangent sections. Addressing roadway departure crashes on rural two lane curves, is a priority for roadway agencies. Previous research has studied roadway and environmental factors and to a limited extent driver factors in lane departure crashes. However, almost no research has addressed the interaction of these three variables and the risk of lane departure. This study utilized data from the SHRP 2 naturalistic driving study and roadway information database to develop a mixed effect logistic regression model to predict the likelihood of a lane encroachment towards the inside of the curve. The model found that direction of the curve, vehicle lane position and amount over the advisory speed limit all increased odds of a lane departure. Two additional models were developed using linear mixed effects models which predicted speed and offset at the point of curvature (PC). The model found the drivers speed and acceleration at 100 m upstream, the curves advisory speed or speed limit, and a driver’s age (>60 years) to affect speed. Offset at the PC was found to be affected by offset at 100 meters upstream, presence of an oncoming vehicle and whether it was dawn/dusk. The results of the speed and offset models can be used in the lane encroachment model to predict the likelihood of a lane departure from 100 meters upstream of the curve.

This paper was sponsored by TRB committee AHB65 Standing Committee on Operational Effects of Geometrics.

01584066

16-5695

English

16p

2016

Transportation Research Board 95th Annual Meeting

Web

Photos; References (20) ; Tables

Automatic data collection systems; Behavior; Driver performance; Drivers; Fatalities; Highway curves; Lane changing; Speeding

Highways; Safety and Human Factors

Transportation Research Board Annual Meeting 2016 Paper #16-5695

TRIS, TRB, ATRI

Jan 12 2016 6:29PM