随着智能航行技术的发展，其在复杂场景应用中面临诸多难题。本文分析了智能航行技术在态势感知、避碰决策和运动控制方面的现状，阐述了交通流密集水域、狭水道及浅水、恶劣天气条件等复杂场景对智能航行技术的挑战，提出了基于多传感器融合的态势感知技术、基于VO–RRT的决策避碰技术以及基于深度强化学习的控制技术等优化策略，以提升智能航行技术在复杂场景下的性能，为智能航行技术的发展提供参考。尽管现有技术取得了一定进展，但仍面临多传感器融合精度不足、VO–RRT算法局部最优性以及深度强化学习训练成本高和可解释性差等问题。为此，本文探讨了引入BEV技术和多模态融合技术以提升态势感知精度，结合深度强化学习与轻量级模型以优化避碰决策，以及通过迁移学习和可解释性算法改进运动控制的潜力。这些改进方向可为智能航行技术的未来发展提供新的思路。
With the development of intelligent navigation technology, it faces numerous challenges in complex - scenario applications. This paper analyzes the current status of intelligent navigation technology in terms of situation awareness, collision avoidance decision - making, and motion control. It expounds on the challenges posed by complex scenarios such as dense traffic flow waters, narrow channels and shallow waters, and adverse weather conditions to intelligent navigation technology. Optimization strategies are proposed, including situation awareness technology based on multi - sensor fusion, decision - making and collision avoidance technology based on VO-RRT, and control technology based on deep reinforcement learning, to improve the performance of intelligent navigation technology in complex scenarios and provide reference for the development of intelligent navigation technology. Although certain progress has been made in existing technologies, there are still problems such as insufficient accuracy in multi-sensor fusion, local optimality of the VO-RRT algorithm, high training costs and poor interpretability of deep reinforcement learning. Therefore, this paper explores the potential of introducing BEV technology and multi-modal fusion technology to improve the accuracy of situation awareness, combining deep reinforcement learning with lightweight models to optimize collision avoidance decisions, and improving motion control through transfer learning and interpretable algorithms. These improvement directions provide new ideas for the future development of intelligent navigation technology.
2026,48(3): 169-175 收稿日期：2025-3-12
DOI：10.3404/j.issn.1672-7649.2026.03.026
分类号：U675
作者简介：陈安(1996-),男,硕士,工程师,研究方向为船舶智能化
参考文献：
[1] 王百勇, 张艳华, 贾俊乾. 基于深度学习理论下电子海图与雷达图像船舶感知信息融合[J]. 现代雷达, 2021, 43(5): 44-50.
WANG B Y, ZHANG Y H, JIA J Q. Ship perception information fusion of electronic chart and radar image based on deep learning theory[J]. Modern Radar, 2021, 43(5): 44-50.
[2] 孙旭, 郑凯, 公丕永等. 智能船舶航行态势感知技术现状[J]. 船舶工程, 2022, 44(4): 14-25.
SUN X, ZHENG K, GONG P Y, et al. Current status of intelligent ship navigation situation awareness technology[J]. Ship Engineering, 2022, 44(4): 14-25.
[3] 赵仓龙. AIS与雷达信号数据融合在船舶避碰系统中的应用[J]. 舰船科学技术, 2022, 44(12): 169-172.
ZHAO C L. Application of AIS and radar signal data fusion in ship collision avoidance system[J]. Ship Science and Technology, 2022, 44(12): 169-172.
[4] 朱旭芳, 黎奥丰, 夏家伟. 基于信息融合的无人艇船舶检测与跟踪[J]. 海军工程大学学报, 2022, 34(5): 20-26.
ZHU X F, LI A F, XIA J W. Unmanned boat ship detection and tracking based on information fusion[J]. Journal of Naval University of Engineering, 2022, 34(5): 20-26.
[5] 周凤杰. 船舶避碰的粒子群-遗传(PSO-GA)的混合优化算法研究[J]. 船舶力学, 2021, 25(7): 909-916.
ZHOU F J. Research on the hybrid optimization algorithm of particle swarm optimization and genetic algorithm for ship collision avoidance[J]. Journal of Ship Mechanics, 2021, 25(7): 909-916.
[6] 李昀哲, 王硕丰, 刘彬, 等. 基于深度强化学习的船舶避碰决策算法[J]. 船舶工程, 2022, 44(10): 118-126.
LI Y Z, WANG S F, LIU B, et al. Ship collision avoidance decision algorithm based on deep reinforcement learning[J]. Ship Engineering, 2022, 44(10): 118-126.
[7] 宁君, 马昊冉, 李伟等. 基于改进RRT算法的船舶路径规划与跟踪控制[J]. 中国航海, 2022, 45(3): 106-112.
NING J, MA H R, LI W, et al. Ship path planning and tracking control based on improved RRT algorithm[J]. China Maritime Affairs, 2022, 45(3): 106-112.
[8] 熊勇, 贺益雄, 黄立文. 基于速度障碍的多船自动避碰控制方法[J]. 中国航海, 2015, 38(3): 46-51.
XIONG Y, HE Y X, HUANG L W. A multi-ship automatic collision avoidance control method based on speed obstacles[J]. China Maritime Affairs, 2015, 38(3): 46-51.
[9] 徐亮, 王欣安, 刘希, 等. 中大型无人船舶自主控制系统架构设计及实现[J]. 中国舰船研究, 2024, 19(S1): 1-9.
XU L, WANG X A, LIU X, et al. Architecture design and implementation of autonomous control system for medium and large-sized unmanned ships[J]. Chinese Journal of Ship Research, 2024, 19(S1): 1-9.
[10] 焦建芳, 胡正中. 无人艇轨迹跟踪的预设性能抗扰控制研究[J]. 信息与控制, 2022, 51(3): 271-279.
JIAO J F, HU Z Z. Research on preset performance robust control for trajectory tracking of unmanned vessels[J]. Information and Control, 2022, 51(3): 271-279.
[11] 沈智鹏, 邹天宇, 王茹. 基于扩张观测器的欠驱动船舶轨迹跟踪低频学习自适应动态面输出反馈控制[J]. 控制理论与应用, 2019, 36(6): 867-876.
SHEN Z P, ZOU T Y, WANG R. Low-frequency learning adaptive dynamic surface output feedback control for underactuated ship trajectory tracking based on extended observer[J]. Control Theory & Applications, 2019, 36(6): 867-876.
