Robust lane recognition embedded in a real-time driver assistance system

We developed a fast and robust approach for automatic lane detection as part of a real-time driver assistance system. Two different algorithms to extract measurement points are used to detect not only marked but unmarked lane borders as well. Different road types as well as various traffic situations and illumination changes require great care on robustness and reliability. Obstacle information computed by another module in this system helps to increase robustness. The algorithm was extended to track two directly neighboured lanes. Additionally, the distribution of the measurement points is used to classify the marking line types. The system has been integrated into two experimental vehicles and tested with a large data set. It performed very well under different traffic situations and weather conditions.