Abstract
This paper deals with a hybrid lane-following system that has both driver assistance and automatic lane-following functions. Discussions in this paper focus on two major topics. One is the algorithm for lane-marker image recognition using a CCD camera and its verification on an experimental vehicle. The algorithm features that it dispenses with state variables on vehicle motion or attitude angles of the camera, nor it requires any high-performance computer. The other is the evaluation of the proposed system's capability to reduce driver's workload in a lane-following task by conducting a simulator test on drivers' lane-following control with secondary task.