Real-Time Long-Range Lane Detection and Tracking for Intelligent Vehicle

Abstract

This paper presents a real-time long-range lane detection and tracking approach to meet the requirements of the high-speed intelligent vehicles running on highway roads. Based on a linear-parabolic two-lane highway road model and a novel strong lane marking feature named Lane Marking Segmentation, the maximal lane detection distance of this approach is up to 120 meters. Then the lane lines are selected and tracked by estimating the ego vehicle lateral offset with a Kalman filter. Experiment results with test dataset extracted from real traffic scenes on highway roads show that the approaches proposed in this paper can achieve a high detection rate with a low time cost.