Abstract: Preview information is obtained by processing data from image data of a CCD camera, and it is treated as a known disturbance. Then, models for a vehicle and a target road path, typically consisting of DARMA (deterministic auto-regressive moving average) models are prepared so as to give a relationship between a front wheel steering angle command and a lateral deviation of the vehicle. An extended generalized predictive control theory is applied to these models so as to compensate for tracking errors that may arise as a result of the known disturbance. Thus, a favorable tracking capability can be achieved with a relatively simple structure, by executing a prediction control which requires only such variable that can be measured both easily and accurately.