道路边界识别是车辆基于道路区域或边界信息自动导航的首要问题．根据驾驶员视觉处理经验和车辆运动轨迹方程，总结出道路边界识别的先验知识与预测知识，在所选取的二次曲线道路形状模型基础上,分别利用边界识别算法和跟踪识别算法得到车辆起步和稳定行驶过程中的实时道路边界信息，为车辆控制器计算出位置偏差和方向偏差两个参数，以实现车辆基于前向单目视觉的自主导航．通过对实际路面试验结果的分析，证明了该方法的有效性和准确性．

Road edge detection and tracking is a key technique for auto navigation of vehicle based on the information of road edge or road region. This paper presents pre knowledge based on human visual experience and prediction knowledge based on the kinematics equation of vehicle. With the selected road shape model, road edge detection and following for intelligent vehicle was executed by the combination of two modules. The road edge detection module is used to get the information about road edge when vehicle being started with the pre knowledge, while the edge following module is started to predict the position of road edge when vehicle running stably with the prediction knowledge, then two control equation parameters location and orientation deviation are calculated for auto navigation of vehicle based on preview unique visual. In the concluding, this paper analyzed the result of the real road tracking experimentation, which proved the effective and precise of the presented method.