基于因子图的激光SLAM模型优化算法  

Factor Graph-based Laser SLAM Model Optimization Algorithm

作　　者：相福磊 彭富明[1] 方斌[1] 张子祥 张少杰 何浩天 XIANG Fulei;PENG Fuming;FANG Bin;ZHANG Zixiang;ZHANG Shaojie;HE Haotian(School of Automation,Nanjing University of Science and Technology,Nanjing 210094,China)

机构地区：[1]南京理工大学自动化学院,江苏南京210094

出　　处：《机械制造与自动化》2024年第5期167-170,208,共5页Machine Building & Automation

基　　金：国家重点研发项目(2021YFE0194600);江苏省科技计划项目(BZ2023023)。

摘　　要：针对激光SLAM点云建图模型研究,设计一种因子图优化的SLAM模型方案。在前端模型中,激光里程计通过引入IESKF实现IMU与雷达点云数据的紧耦合构建IESKF-LIO。在SLAM后端模型中,为提高SLAM的鲁棒性和实时精度,构建多种因子进行位姿约束与补偿优化,同时在因子图融合过程中提出关键帧和增量式平滑建图,减少模型运算负担。在KITTI数据集中进行建图实验,实验结果验证了该模型较传统SLAM模型轨迹误差更低,建图效果更好。For the study of laser SLAM point cloud mapping model,a factor graph optimized SLAM model scheme is designed.In the front-end module,the laser odometry realizes the tight coupling of IMU and radar point cloud data by introducing IESKF,thus IESKF-LIO being constructed.In the SLAM back-end model,a variety of factors are built for pose constraints and compensation optimization in order to improve the robustness and real-time accuracy of SLAM,meanwhile,key frames and incremental smooth mapping are proposed during the factor graph fusion process to reduce the computational burden of the model.Through the mapping experiments in KITTI data sets,the experimental results verify that the designed model has lower trajectory errors and better mapping effects than the traditional SLAM onel.

关 键 词：SLAM 因子图 IESKF 回环检测 

分 类 号：TP391.9[自动化与计算机技术—计算机应用技术]