Abstract: The active vision method was used in the control of intelligent vehicles on urban environment roads to expand the field of view of onboard camera. The angle of the pan&tilt unit is set according to the relative position between the lane and the vehicle to keep the lane in the middle of the image. The corresponding relationship between the vehicle coordinate system and image coordinate system was analyzed and calculated in real time. The conclusion that the active vision system has a higher roadtracking accuracy and less imagelost rate is shown in the comparison experiments between the active vision system and the traditional vision system.