The Advanced Driving Assistance System has been widely applied on various kinds of vehicles, since it can provide timely protection and burden relief to the passengers and drivers. Multiple different sensors are adopted to realize specific functions, while millimeter-wave radar and vision sensor are one of the most common sensor combinations. The scheme of sensor data fusion for this combination becomes necessary to contribute with more accurate and complementary detection results. A data fusion model designed for the pair of radar and vision is proposed with a three-level hierarchical structure, which includes screening of original sensor data, sensor data correlation, and target track forecast. The concrete fusion scheme applied on each layer is presented in details. The results of simulation and real circumstances test validate and prove the presented sensor data fusion model is efficient.

先进驾驶辅助系统已广泛应用于各类车辆，为乘客和司机提供及时保护和减轻负担。采用多种不同的传感器来实现特定的功能，而毫米波雷达和视觉传感器是最常见的传感器组合之一。这种组合的传感器数据融合方案变得必要，以提供更准确和互补的检测结果。提出了一种为雷达和视觉对设计的数据融合模型，采用三级层次结构，包括原始传感器数据筛选、传感器数据关联和目标轨迹预测。详细介绍了应用于每一层的具体融合方案。