A New Evaluation Method to Quantify Drivers' Lane Keeping Behaviors on Urban Roads

Abstract

Poor lane keeping behaviors may cause problems and 'Level' of lane keeping has received little attention from the researchers to gain a better understanding on degree of untidiness. Unfortunately, previous studies have not clarified the drivers' lane keeping levels. For this reason, 'level of untidy' should be defined and the question of 'in which level' needs to be answered. Paper suggests five levels for quantitative measure of (un)tidiness so that the effects can be assessed for traffic flow models. For this aim, proposed levels are analyzed by considering road, vehicle, and traffic properties in urban roads. An ordered probit regression analysis was performed to investigate relation between keeping level and effective parameters using each vehicle's lateral position data. Results showed that volume and lane width parameters have a positive effect, and vehicle width and lane number variables have negative effect on keeping level as continuous variables of the model.