Vehicle lane departure estimation on urban roads using GIS information

Abstract

The ego vehicle's location is critical for a variety of autonomous functions, the most notable of which are lane departure warning systems (LDWS). LDWS typically detect lane departures in urban environments by utilizing a variety of vision processing or optical scanning techniques. These LDWS have some limitations; for example, they may be impaired by adverse weather conditions or irregular lane markings. Other LDWS employ a global positioning system (GPS) receiver and digital maps with lane-level resolution to increase system efficiency, but at the expense of added complexity and cost. To compensate for camera failures, this paper proposes an alternate system that is not reliant on lane markings and instead uses GPS and IMU data collected from the vehicle's smartphone to locate the vehicle on the road and compare its trajectory to a reference road direction without requiring any digital maps with lane-level resolution. This technique only requires road-level data from a common digital mapping database. Additionally, experiments demonstrate that the proposed system is reliable and can be supplemented with a camera to improve vehicle estimates at the lane level.