报告详情
  1. 活动首页
  2. 报告清单
  3. 报告详情
A Hybrid Algorithm for Loop Closure Detection in LiDAR SLAM Using Global Point Cloud Density and Local Triangular Features
编号：54 访问权限：仅限参会人 更新：2025-11-10 11:28:03 浏览：41次 口头报告

报告开始：2025年11月23日 09:50（Asia/Shanghai）

报告时间：20min

所在会场：[S1] Parallel Session 1 [S1-2] Parallel Session 1-23 AM

暂无文件

摘要
In Simultaneous Localization and Mapping (SLAM) systems, LiDAR odometry inevitably accumulates drift in pose estimates, leading to significant trajectory deviations and inconsistencies in the global map. Loop closure detection plays a crucial role in mitigating this cumulative error through closed-loop optimization strategies. In this paper, we propose a hybrid algorithm for loop closure detection in LiDAR SLAM that leverages both global point cloud density and local triangular features. Specifically, the algorithm first filters candidate keyframes using a ring descriptor based on global point cloud density, followed by fine-grained triangle matching and geometric verification within each candidate frame to identify the optimal loop closure frame. The proposed method was evaluated on the publicly available Park dataset, with experimental results demonstrating a reduction in the RMSE of trajectory error to 0.051 m, representing a 59% improvement over the original odometry results, and outperforming other benchmark algorithms in terms of trajectory error reduction. Additionally, the precision-recall (PR) curve of the proposed algorithm exhibited the widest coverage, achieving a maximum F1 score of 0.92, thereby outperforming existing mainstream loop closure detection algorithms.
关键词
Simultaneous Localization and Mapping,loop closure detection,point cloud density,triangular feature
报告人
Yixiao Chen
Master candidate Southeast University

稿件作者
Yixiao Chen Southeast University
Xuefen Zhu Southeast University
Gangyi Tu Nanjing University of Information Science and technology
Chuan Zhang Southeast University
Ang Li Southeast University
发表评论
验证码 看不清楚，更换一张
全部评论
重要日期

  • 会议日期

    11月21日

    2025

    至

    11月23日

    2025

  • 10月20日 2025

    初稿截稿日期

  • 11月23日 2025

    注册截止日期

主办单位
IEEE Instrumentation and Measurement Society
South China University of Technology
承办单位
South China University of Technology
联系方式
登录查看完整联系方式
历届会议
移动端
在手机上打开
小程序
打开微信小程序
客服
扫码或点此咨询