基于扫描上下文优化的紧耦合激光SLAM方法
CSTR:
作者:
作者单位:

1. 长沙理工大学 电气与信息工程学院,长沙 410114;2. 湖南大学 机器人学院,长沙 410082

作者简介:

通讯作者:

E-mail: zhanghuihby@126.com.

中图分类号:

U463.6;TN958.98

基金项目:

科技创新2030-“新一代人工智能”重大项目(2021ZD0114503);国家自然科学基金重大研究计划项目(92148204);国家自然科学基金项目(62027810,61971071);湖南省科技创新领军人才项目(2022RC3063);湖南省杰出青年科学基金项目(2021JJ10025);湖南省重点研发计划项目(2021GK4011,2022GK2011);长沙科技重大项目(KH2003026);机器人国家重点实验室联合开放基金项目(2021-KF-22-17);中国高校产学研创新基金项目(2020HYA06006).


Tightly coupled laser SLAM method base on scan context optimization
Author:
Affiliation:

1. School of Electrical and Information Engineering,Changsha University of Science and Technology,Changsha 410114,China;2. School of Robotics,Hunan University,Changsha 410082,China

Fund Project:

  • 摘要
  • |
  • 图/表
  • |
  • 访问统计
  • |
  • 参考文献
  • |
  • 相似文献
  • |
  • 引证文献
  • |
  • 资源附件
  • |
  • 文章评论
    摘要:

    高精度地图与定位是无人车实现自主导航作业的重要前提,针对现有松耦合融合方法对观测信息利用不充分,以及传统闭环检测方法匹配精度低等问题,提出一种基于扫描上下文优化的紧耦合激光SLAM方法-----optimized SC-LIO-SAM.首先,IMU通过预积分对点云进行去偏校正,同时为激光里程计提供初始位姿估计;激光里程计通过滑动窗口的方法将当前帧的特征点云与局部地图匹配,随后基于扫描上下文的方法对特征点云进行编码生成点云描述符,实现高效的闭环检测;基于LIO-SAM的框架,将IMU预积分因子、激光里程计因子、GPS因子以及闭环因子插入全局因子图中,通过基于贝叶斯树的增量平滑优化算法对全局点云优化更新.为了验证所提方法的有效性,采用KITTI数据集评估optimised SC-LIO-SAM的性能,并与LOAM、LEGO-LOAM以及LIO-SAM对比,结果表明,optimised SC-LIO-SAM相比于LOAM、LEGO-LOAM以及LIO-SAM等算法,定位精度显著提升.最后将算法应用在开源数据集中,验证了optimised SC-LIO-SAM能够构建全局一致的地图.

    Abstract:

    High-precision mapping and positioning is an important prerequisite for autonomous navigation of unmanned vehicles, and in view of the insufficient utilization of observation information by existing loosely coupled fusion methods and the low matching accuracy of traditional closed-loop detection methods, a tightly coupled laser SLAM method based on scan context optimization, named optimized SC-LIO-SAM, is proposed. First, the IMU debiases the point cloud through pre-integration and provides initial pose estimation for the laser odometer. The laser odometer matches the feature point cloud of the current frame with the local map by sliding the window. Then, based on the scan context, the feature cloud is encoded to generate point cloud descriptors to achieve efficient closed-loop detection. Based on the LIO-SAM framework, the IMU pre-integration factor, laser odometry factor, GPS factor and closed-loop factor are inserted into the global factor gragh, lastly the global node optimization is updated by using Bayesian tree. In order to verify the effctiveness of the proposed method, the performance of the optimized SC-LIO-SAM is evaluated by KITTI dataset and compared with LOAM, LEGO-LOAM and LIO-SAM, and the experimental results show that the positioning accuracy of the optimized SC-LIO-SAM is significantly improved compared with LOAM, LEGO-LOAM, LIO-SAM and the classical algorithm. Finally, the algorithm is applied to the open-source dataset to prove that the optimized SC-LIO-SAM can build a globally consistent map.

    参考文献
    相似文献
    引证文献
引用本文

汪湘川,张辉,陈波,等.基于扫描上下文优化的紧耦合激光SLAM方法[J].控制与决策,2024,39(10):3234-3242

复制
分享
文章指标
  • 点击次数:
  • 下载次数:
  • HTML阅读次数:
  • 引用次数:
历史
  • 收稿日期:
  • 最后修改日期:
  • 录用日期:
  • 在线发布日期: 2024-08-29
  • 出版日期: 2024-10-20
文章二维码