路面机动目标的融合跟踪

Fusion Tracking for Maneuvering Targets

针对自主车辆的机动目标跟踪问题提出了一个利用雷达和图像传感器的融合跟踪方法。采用动力学模型对车辆运动进行描述以适应车辆机动的情形。利用雷达的距离信息建立视觉窗口,利用三维线框模型从窗口图像中获取车辆转向角,然后通过数据融合技术,充分利用雷达与图像传感的量测信息,提高对机动目标的状态估计精度。三维仿真实验证明,与单纯基于雷达量测的跟踪方法相比,该融合方法大大提高了对机动目标的跟踪精度。

A new fusion tracking system for maneuvering vehicle based on radar and image sensors is presented. Dynamic model is adopted for vehicle moving description, which satisfies the complicated mobile cases. Vision tracking window is established using range information from radar, and orientation data of vehicle is obtained from window image by 3-D wire|frame model method. Then data fusion technology is adopted to make full use of radar and vision sensor's observation, which helps to improve state estimation accuracy of mobile targets. 3-D simulation has confirmed that the proposed fusion method can track maneuvering targets more accurately compared with radar|based tracking.