Harnessing retroreflective transportation infrastructure for intelligent vehicle positioning

Abstract

The popularity of intelligent transportation system (ITS) applications urges the development of vehicle positioning system that satisfies the availability, precision and accuracy requirements. Lane level (the exact lane on the road) positioning or even in-lane level (the exact location within the lane) positioning are expected in all driving environments at all times. In addition to the existing vehicle positioning technologies, such as Global Positioning System (GPS), terrestrial radio (e.g., cell phone, AM, FM and digital TV) and feature-based sensors (e.g., vision, LiDAR and RADAR), we propose a complementary vehicle positioning system based on ubiquitous retroreflective transportation infrastructures. Low-cost photodiodes are mounted on the headlights to sense the retroreflected optical signals from on-road safety device, such as raised pavement marker (RPM). A liquid crystal display (LCD) shutter is installed on the RPM to distinguish the retroreflected light signals from different RPMs and from the environmental reflection. By sensing the angle-of-arrival of the retroreflected light beam from the RPMs (i.e., landmarks), the vehicle realizes lane level positioning based on triangulation. Advanced positioning solution is proposed by integrating dead reckoning and Kalman filter to advance the lane level positioning to in-lane level positioning. Extensive experimental results reveal that the proposed vehicle positioning system achieves sub-meter level location accuracy.