A Sensor Fusion Digital-Map System for Driver Assistance

A traffic situation is getting more complex in urban areas. Various safety systems of an automobile have been developed but fatal and serious accidents still can be made by driver's faults or distractions. The system supporting extend of driver's recognition area is going to be an important part of future intelligent vehicles in order to prevent accidents.

In this paper we propose sensor fusion system based on a digital-map for driver assistance. The accurate localization of a host vehicle is achieved by a stereo vision sensor and a digital-map using polygon matching algorithm in urban area. A single-row laser scanner is used for tracking multiple moving objects. The coordinate transformation from sensor frame to global frame is performed to visualize the moving objects on a digital-map.

An experiment was conducted in an urban canyon where the GPS signals are frequently interrupted. Four cameras were installed in the left and right of the vehicle to get images of landmarks and horizontal laser scanner was mounted to collect scan data.