水域环境中障碍物分布呈非结构化，受自然因素干扰，生成的路径存在曲率突变、转折点密集的问题，导致无人船路径平滑性差、避障能力弱，故提出基于深度学习的无人船动态路径规划方法。通过栅格法离散化水域，构建无人船航行动态环境模型；结合深度强化学习的全局决策与动态窗口法的局部避障能力，实现路径动态规划，经速度空间约束与评价函数初步优化轨迹；通过贪心剪枝去除冗余节点简化路径，用三次B样条曲线平滑关键点。实验表明，该方法有效提升了路径规划的平滑性与安全性。
The distribution of obstacles in the water environment is unstructured and affected by natural factors, resulting in paths with abrupt curvature changes and dense turning points, leading to poor smoothness and weak obstacle avoidance ability of unmanned ships. Therefore, this study proposes a dynamic path planning method based on deep learning. Discretize the water area using the grid method and construct a dynamic environment model for unmanned ship navigation. By combining the global decision-making of deep reinforcement learning with the local obstacle avoidance capability of dynamic window method, dynamic path planning is achieved, and the trajectory is preliminarily optimized through velocity space constraints and evaluation functions. Simplify the path by removing redundant nodes through greedy pruning, and smooth the key points with cubic B-spline curves. The experiment shows that this method effectively improves the smoothness and security of path planning.
2025,47(24): 196-200 收稿日期：2025-3-8
DOI：10.3404/j.issn.1672-7649.2025.24.032
分类号：U674
作者简介：刘芳(1990-),女,硕士,实验师,研究方向为深度学习
参考文献：
[1] 李军涛, 侯星星, 茆俊亚, 等. 时变海流干扰下深远海渔业无人船多目标路径规划[J]. 上海海洋大学学报, 2023, 32(5): 1090-1098.
LI J T, HOU X X, MAO J Y, et al. Multi-objective path planning for unmanned vessels in deep-sea fisheries under time-varying current disturbance[J]. Journal of Shanghai Ocean University, 2023, 32(5): 1090-1098.
[2] 乔珍, 尹传忠, 仇鑫. 基于改进蚁群算法的长航程无人船路径规划[J]. 计算机工程与科学, 2024, 46(10): 1835-1842.
QIAO Z, YIN C Z, QIU X. Path planning of long-range unmanned ship based on improved ant colony algorithm[J]. Computer Engineering & Science, 2024, 46(10): 1835-1842.
[3] 林潭奇, 李存荣, 刘帅文. 基于改进A*算法的内陆湖无人船路径规划[J]. 大连海事大学学报, 2024, 50(3): 87-96.
LIN T Q, LI C R, LIU S W. Path planning of unmanned ships in inland lakes based on improved A* algorithm[J]. Journal of Dalian Maritime University, 2024, 50(3): 87-96.
[4] XIAO Q, JIANG L, WANG M, et al. An improved distributed sampling ppo algorithm based on beta policy for continuous global path planning scheme[J]. Sensors, 2023, 23(13): 6101.
[5] WU H, WANG F, MEI X, et al. A novel fuzzy control path planning algorithm for intelligent ship based on scale factors[J]. The Journal of Supercomputing, 2024, 80(1): 202-225.
[6] 姜龙腾, 迟瑞娟, 马悦琦, 等. 基于栅格法的农业机器人路径规划方法研究[J]. 农机化研究, 2024, 46(6): 19-24.
JIANG L T, CHI R J, MA Y Q, et al. Research on path planning method of agricultural robot based on grid method[J]. Journal of Agricultural Mechanization Research, 2024, 46(6): 19-24.
[7] 江坤颐, 孙世平, 蒋丙栋, 等. 基于导航雷达回波视频数据的占据栅格地图构建方法[J]. 中国舰船研究, 2025, 20(1): 96-106.
JIANG K Y, SUN S P, JIANG B D, et al. Method for constructing occupancy grid maps based on navigation radar echo video data[J]. Chinese Journal of Ship Research, 2025, 20(1): 96-106.
[8] 张伟龙, 单梁, 常路, 等. 基于改进DWA的多无人水面艇分布式避碰算法[J]. 控制与决策, 2023, 38(4): 951-962.
ZHANG W L, SHAN L, CHANG L, et al. Distributed collision avoidance algorithm for multiple unmanned surface vessels based on improved DWA[J]. Control and Decision, 2023, 38(4): 951-962.
[9] 张啸天, 陈熙源. 基于IRRT*和DWA的无人艇混合路径规划方法[J]. 传感技术学报, 2022, 35(11): 1469-1474.
ZHANG X T, CHEN X Y. Path planning method for unmanned surface vehicle based on IRRT*and DWA[J]. Chinese Journal of Sensors and Actuators, 2022, 35(11): 1469-1474.
[10] 杜传胜, 高焕兵, 侯宇翔, 等. 同根双向扩展的贪心RRT路径规划算法[J]. 计算机工程与应用, 2023, 59(21): 312-318.
DU C S, GAO H B, HOU Y X, et al. Greedy RRT path planning algorithm with same root bidirectional extension[J]. Computer Engineering and Applications, 2023, 59(21): 312-318.
