为提升无人船的海洋环境感知能力，助力路径规划、避障决策等工作顺利开展，研究激光雷达数据驱动下无人船障碍物感知定位方法。通过无人船搭载激光雷达采集航行环境机激光雷达点云数据，经坐标转换至无人船坐标后，结合姿态仪数据进行坐标优化，并通过坐标映射构建变尺寸栅格地图，运用基于DBSCAN算法的数据挖掘方法去除海杂波干扰以精准感知障碍物，通过多假设跟踪模型和卡尔曼滤波器实现障碍物定位与动态跟踪。结果表明，该方法能有效利用激光雷达数据感知障碍物信息，并准确跟踪障碍物位置变化，具备良好稳定性。
To enhance the marine environment perception ability of unmanned vessels and facilitate the smooth progress of tasks such as path planning and obstacle avoidance decision-making, the obstacle perception and positioning method of unmanned vessels driven by lidar data is studied. Collecting navigation environment machine LiDAR point cloud data through unmanned ships carrying LiDAR, converting coordinates to unmanned ship coordinates, optimizing coordinates with attitude instrument data, and constructing variable size grid maps through coordinate mapping. Using data mining methods based on DBSCAN algorithm to remove sea clutter interference and accurately perceive obstacles, obstacle localization and dynamic tracking are achieved through multi hypothesis tracking model and Kalman filter. The results show that this method can effectively utilize LiDAR data to perceive obstacle information and accurately track changes in obstacle positions, with good stability.
2025,47(12): 55-58 收稿日期：2024-7-18
DOI：10.3404/j.issn.1672-7649.2025.12.011
分类号：U664.82
作者简介：赵越(1977-),男,副教授,研究方向为航海技术及智能船舶控制
参考文献：
[1] 沙如月. 无人船与传统船舶在海洋监测中的协同作业模式 [J]. 船舶物资与市场, 2024, 32 (9): 124-126.
[2] 周金涛, 高迪驹, 刘志全. 基于全景视觉的无人船水面障碍物检测方法[J]. 计算机工程, 2024, 50(2): 113-121.
[3] 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.
[4] 董文博, 周利, 刘仁伟, 等. 基于边界框重叠最大化的海上单目标实时跟踪方法[J]. 中国造船, 2023, 64(1): 205-214.
[5] 倪桦, 关巍, 张显库. 基于激光雷达与摄像头的无人船目标感知与测距[J]. 船舶工程, 2022, 44(9): 107-113.
[6] 陈卓, 王飞, 陈奕宏, 等. 基于激光雷达的无人艇海上目标检测与跟踪方法研究[J]. 中国造船, 2022, 63(6): 264-272.
