Lane detection for a steering assistance system

Abstract

In this paper, a lane detection algorithm for a steering assistance system (SAS) is introduced. Our aim is to develop a lane detection sensor that is sufficiently robust to enable it to be applied to the SAS without the need for application specific hardware. For this purpose, we simplify the process that groups detected edge points into lines, using a road surface with low resolution. Additionally, we use a method to select the correct lane boundary lines from multiple candidates using pattern matching for scenes where lane boundaries are complex. Our algorithm has been implemented on simple hardware consisting of a CMOS imager and two microprocessors.