Abstract: Aiming at the problems  of low positioning accuracy and robustness, as well as poor precision and stability in mapping for pasture inspection robots, a novel LOMSLAM algorithm based on LiDAR ranging and mapping technology and improved LOMSLAM algorithm is proposed. This algorithm is derived from an enhanced LOAMSLAM algorithm, which integrates laser range finding and surveying technology. LOMSLAM decomposes SLAM into two separate processes such as motion estimation and map construction. By leveraging the high precision of laser range finding and surveying technology, LOMSLAM achieves simultaneous robot localization and map building, thereby enhancing the accuracy, robustness, and stability of both positioning and mapping. LOMSLAM was installed on a designed inspection robot with Mecanum wheel structure for test verification. The results showed that in pose estimation tests, LOMSLAM significantly outperformed other methods in terms of relative pose error (RPE) and absolute trajectory error (ATE), with RMSE values of just 7.28m and 2.23m, respectively, which were lower than the comparative algorithms. In the positioning and mapping tests, with the inspection robot moving at speeds of 0.2 m/s, 0.5 m/s, and 1 m/s, the positioning errors of LOMSLAM were only 0.12 m, 1 m, and 1.2 m, respectively, demonstrating better positioning accuracy and robustness compared to the comparative algorithms.

Key words: inspection robots, LiDAR, improved SLAM, pastoral environment, positioning and mapping

摘要： 针对牧场巡检机器人定位精度和鲁棒性低、建图精度和稳定性差的问题，提出一种基于激光雷达测距和测绘技术与改进LOAMSLAM算法的LOMSLAM算法。LOMSLAM算法在LOAMSLAM算法的基础上将SLAM分解为运动估计和地图构建两个过程，利用激光雷达的高精度测距和测绘技术，实现同时进行机器人的定位和地图构建，从而提高定位与建图的精度，提高鲁棒性和稳定性。将LOMSLAM搭载在麦轮结构的巡检机器人上进行试验验证。结果表明：在位姿估计试验中，LOMSLAM算法的绝对轨迹误差（ATE）和相对位姿误差（RPE）的RMSE值分别仅为7.28m和2.23m，均低于对比算法。在定位与建图试验中，当巡检机器人分别以0.2 m/s、0.5 m/s、1 m/s的速度运动时，LOMSLAM的定位误差分别为0.12 m、1 m、1.2 m，具有更好的定位精度和稳健性。

关键词: 巡检机器人, 激光雷达, 改进SLAM, 牧场环境, 定位与建图