摘要： 光伏河蟹养殖池塘环境中，受到太阳能电池板遮挡，致使无人作业船卫星定位系统的精度过低。提出一种基于激光惯性的无人作业船定位方法。首先，对雷达点云数据进行预处理，滤除扰动，降低规模，有效提升了数据的准确性；其次，改进Hector-Slam的定位流程，采用非线性拟合确定障碍物中心，再利用高斯虚化对障碍物中心进行虚化处理，实现地图连续化。同时，采用梯度下降法进行地图匹配，定位结果的抖动明显改善，精度明显提升；最后，采用卡尔曼滤波器对雷达位姿与IMU位姿进行融合，得到更加精确的位姿信息。经样机试验，采用激光惯性融合定位方法得到的定位偏差均值，相比Hector算法定位缩小了46%。研究结果表明，该激光惯性融合定位方法，能代替卫星定位，满足实际生产需求。
关键词：无人作业船；导航定位；2D激光SLAM；数据融合；卡尔曼滤波

关键词: 无人作业船, 导航定位, 2D激光SLAM, 数据融合, 卡尔曼滤波

Abstract: In the photovoltaic river crab breeding pond environment, solar panel obstruction significantly reduces the accuracy of the unmanned operation ship's satellite positioning system. To address this, a laser - inertia - based unmanned operation ship positioning method is proposed, considering the pond's unique conditions. This method improves the Hector - Slam positioning process. First, LiDAR point cloud data undergoes preprocessing to filter disturbances and reduce data size, enhancing accuracy. Then, the map continuity method in Hector - Slam is enhanced by using nonlinear fitting to identify obstacle centers, followed by Gaussian blurring to ensure map continuity, creating a smoother reference map for matching. Next, the map - matching process in Hector - Slam is improved by replacing the Gaussian Newton method with gradient descent, yielding more precise results. Finally, a Kalman filter integrates radar and IMU poses, combining position and heading angle information for improved positioning accuracy. Experimental results show the laser inertial fusion positioning method reduces average positioning deviation by 46% compared to the Hector algorithm. Unlike satellite positioning, which fails to meet the accuracy requirements in photovoltaic ponds, our laser - based method ensures precise positioning. It also outperforms visual schemes in accuracy under disturbances and low - light conditions. Moreover, compared to high - cost 3D laser solutions that are impractical for agricultural production, our cost - effective laser method offers significant advantages. Thus, this laser inertial fusion positioning method can replace satellite positioning, effectively meeting practical production needs.

Key words:  , unmanned operation ship, navigation and positioning, 2D Laser SLAM, data fusion, kalman filter