Unmanned surface vehicle is increasingly becoming a research hotspot, which can be used in a variety of civil and military missions. However, compared with the relative maturity of other technologies, the sensing technology of unmanned surface vehicles is relatively weak. Taking "WAM-V-USV" as the research platform, this paper is mainly focus on the detection and tracking methods of moving objects with unmanned surface vehicles. This paper introduces the environment sensing system of unmanned vehicle, water surface image preprocessing, water antenna detection based on SVM, and the method of water surface object detection and tracking based on improved YOLOV3. The simulation results show that the proposed method can effectively improve the accuracy of moving object detection and tracking. Through the practical application in the Songhua River and the US Unmanned Surface Vehicles Open, it is proved that the algorithm has a good detection and tracking effect and meets the real-time requirements. Practice has proved that the object detection and tracking method based on deep learning greatly improves the perception ability and self-security of the unmanned surface vehicles.

无人水面飞行器正日益成为研究热点，可用于各种民用和军事任务。但是，与其他技术的相对成熟相比，无人水面车辆的传感技术相对薄弱。本文以“ WAM-V-USV”为研究平台，重点研究了无人水面载具的运动物体检测与跟踪方法。介绍了无人机环境传感系统，水面图像预处理，基于支持向量机的水天线检测以及基于改进型YOLOV3的水面目标检测与跟踪方法。仿真结果表明，该方法可以有效提高运动目标的检测和跟踪精度。通过在松花江和美国无人水面公开赛的实际应用，证明该算法具有良好的检测和跟踪效果，满足实时性要求。实践证明，基于深度学习的目标检测与跟踪方法极大地提高了无人机的感知能力和自我安全性。