摘要
激光-惯性同时定位与建图(SLAM)在自动驾驶与机器人导航中具有广泛应用,但传统点云配准方法在处理大规模室外数据时存在效率低和精度不足的问题。为此,提出一种基于自适应面元配准的激光-惯性SLAM方法。通过动态调整面元大小以适应不同环境特征,并利用自适应面元构建地图和配准,优化了配准过程。实验结果表明,与基于点云地图配准的方法相比,所提方法显著提高了配准精度并增强了系统鲁棒性。在KITTI数据集上,与Le GO-LOAM和LIO-SAM相比,平均定位误差分别降低约47.8%和39.1%;在实测实验中,相较于Le GO-LOAM、LIO-SAM及Fast-LIO2,平均定位误差分别降低约45.9%、34.4%和56.3%。
LiDAR-inertial simultaneous localization and mapping(SLAM)is widely used in autonomous driving and robot navigation.Traditional point cloud-based registration methods face challenges in efficiency and accuracy when processing large-scale outdoor data.To address this,an adaptive surfel based registration method for LiDAR-inertial SLAM is proposed.By dynamically adjusting the size of surfel to adapt to different environmental features,the method optimizes the registration process using adaptive surfel for mapping and registration.Experimental results show that compared to point cloud-based registration methods,the proposed method significantly improves registration accuracy and enhances system robustness.On the KITTI dataset,the average positioning error is reduced by 47.8%and 39.1%compared to LeGO-LOAM and LIO-SAM,respectively.In real-world experiments,the average positioning error is reduced by approximately 45.9%,34.4%and 56.3%compared to LeGO-LOAM,LIO-SAM,and Fast-LIO2,respectively.
作者
徐晓苏
张家赫
XU Xiaosu;ZHANG Jiahe(Key Laboratory of Micro Inertial Instrument and Navigation Technology,Ministry of Education,Nanjing 210096,China;Institute of Instrument Science and Engineering,Southeast University,Nanjing 210096,China)
出处
《中国惯性技术学报》
北大核心
2025年第6期587-595,共9页
Journal of Chinese Inertial Technology
基金
装备重大基础研究项目(51405-02A03)
国家自然科学基金项目(61921004)。
关键词
激光-惯性SLAM
点云配准
面元地图
LiDAR-inertial simultaneous localization and mapping
point cloud registration
surfel map
