摘要
针对智能船舶的自主航行障碍物视觉快速检测与测距需求,提出一种基于深度学习的智能船舶轻量化水面障碍物视觉检测与测距方法。从障碍物检测速度和计算量的角度出发,该方法可提升智能船舶环境感知能力。首先,针对障碍物检测问题,在Yolov4检测模型的框架下,构建基于MobileNet特征提取网络的DIS-Yolo水面障碍物检测模型,实现模型网络结构的轻量化改进。其次,针对障碍物测距问题,基于所构建的障碍物检测模型和COMS成像模型,提出水面障碍物测距机制,实现水面障碍物的高精度测距。最后,通过模拟实验验证所改进模型的有效性与测距函数的精确度。所提出的方法可提升智能船舶的航行安全性,同时可为智能船舶环境感知需求提供新的思路。
Aiming at the requirements of rapid visual detection and ranging of obstacles for autonomous navigation of intelligent ships,a deep learning-based lightweight water surface obstacle visual detection and ranging method is proposed for intelligent ships.From the perspective of obstacle detection speed and computational complexity,this method can enhance the environmental perception capabilities of intelligent ships.For the obstacle detection problem,the DIS-Yolo water surface obstacle detection model based on the MobileNet feature extraction network is constructed within the framework of the Yolov4 detection model,achieving a lightweight improvement of the model network structure.For the obstacle ranging problem,a water surface obstacle ranging mechanism is proposed based on the constructed obstacle detection model and COMS imaging model,realizing high-precision ranging of water surface obstacles.Finally,the effectiveness of the improved model and the accuracy of the ranging function are verified through simulated experiments.The proposed method can improve the navigation safety of intelligent ships and provide new ideas for the environmental perception needs of intelligent ships.
作者
朱凡
潘宝峰
马勇
祝贵兵
吴中岱
ZHU Fan;PAN Baofeng;MA Yong;ZHU Guibing;Wu Zhongdai(School of Naval Architecture and Maritime,Zhejiang Ocean University,Zhoushan 316022,China;National Key Laboratory of Waterway Traffic Control,Wuhan University of Technology,Wuhan 430063,China;School of Navigation,Wuhan University of Technology,Wuhan 430063,China;National Engineering and Technology Research Center of Water Transport Safety,Wuhan University of Technology,Wuhan 430063,China;Cosco Marine Technology Co.LTD,Shanghai 200135,China;Cosco Marine Technology Co Academician Workstation,Shanghai 200135,China.)
出处
《无人系统技术》
2023年第5期17-27,共11页
Unmanned Systems Technology
