Robust lane detection algorithm based on triangular lane model

Recently, various technologies and sciences are developed for fourth industrial revolution which includes artificial intelligence, robotics, internet of things, 3-D printing, and autonomous transportation. The autonomous transportation is one of the fourth industrial technology, and has been studying for safety driving. Accurate lane detection and lane departure warning system is very important for autonomous transportation. However, conventional methods have some problems of applying real-driving such as extracting miss lanes under cloudy weather and vehicle disturbance situations. In order to solve the real driving problems, we propose a new lane detection technique for lane departure and forward collision warning system using a single in-vehicle camera. The proposed method consists of triangular lane model, feature points extraction method. In the near field, a triangular lane model is used to approximate a pair of lane boundaries. Subsequently, feature points extraction method based on hyperbola curve model is applied to obtain lane curvature in the far field. Mathematical B-spline applied to feature points for curved lane fitting. Simulation results show that the proposed lane detection and tracking method has good performance.