A lane detection system is an important component of many intelligent transportation systems. In this paper, we present a robust and real-time lane tracking algorithm for any curved road. We use Non Uniform B-Spline (NUBS) interpolation method to construct the boundaries road lane. First, the Canny filter is employed to obtain edge map. Second, we apply parallel thinning algorithm into edge map to get the skeleton image. Once skeleton image is obtained, we introduce a new algorithm based on Vector Lane Concept to extract feature edge points that play the role as control points for NUBS interpolation process to represent left and right lane, separated. Finally, we estimate the curvature of left and right lane boundary for lane tracking. The result of this method can be used for Autonomous Guided Vehicles (AGV) system application.