Acta Optica Sinica, Volume. 44, Issue 20, 2015001(2024)

Laser Simultaneous Localization and Mapping Algorithm Based on Optimized Ground Segmentation and Closed-Loop Detection

Zhaoqiang Li1,2, Yue Zhang1,2、*, Fuli Xiong1, and Huijie Su1,2
Author Affiliations
  • 1School of Information and Control Engineering, Xi’an University of Architecture and Technology, Xi’an 710055, Shaanxi , China
  • 2Xi’an Key Laboratory of Smart Industry Perception, Computing and Decision-making, Xi’an University of Architecture and Technology, Xi’an 710055, Shaanxi , China
  • show less
    Figures & Tables(14)
    Frame diagram of laser SLAM system
    Modeling diagram. (a) Traditional modeling method; (b) concentric zone model
    Statistical diagram of point cloud distribution range
    Filtered plot of ground segmentation on KITTI sequence 00
    LiDAR odometry framework based on ground segmentation
    Feature extraction process in KITTI sequence 00. (a) Original point cloud; (b) planar features of vertical and horizontal parts of Fpk and Fgk; (c) vertically distributed edge features in Fek; (d) spherical features in Fsk
    Loop-back detection based on global quasiregistration
    Comparison of trajectories on KITTI dataset. (a1), (b1), and (c1) are comparison of trajectories of sequences 00, 02, and 10; (a2), (b2), and (c2) are detail comparison corresponding to box selections in (a1), (b1), and (c1)
    Outdoor data acquisition ground robot
    Mapping comparison of M2DGR dataset. (a), (b), and (c) are 3D top view map, 3D local map, and 3D local map (z-axis perspective) corresponding to LOAM algorithm; (d), (e), and (f) are 3D top view map, 3D local map, and 3D local map (z-axis perspective) corresponding to LeGO-LOAM algorithm; (g), (h), and (i) are 3D top view map, 3D local map, and 3D local map (z-axis perspective) corresponding to FAST-LIO algorithm; (j), (k), and (l) are 3D top view map, 3D local map, and 3D local map (z-axis perspective) corresponding to proposed algorithm
    3D mapping of underground mine slope path based on proposed algorithm
    • Table 1. Absolute pose error of each sequence for different algorithms

      View table

      Table 1. Absolute pose error of each sequence for different algorithms

      SequenceLoopAlgorithmType of error /m
      RMSEMeanMedianSTD
      00LOAM18.5720.5119.2310.56
      LeGO-LOAM19.1517.8217.827.01
      FAST-LIO8.655.736.233.21
      Proposed3.272.372.691.70
      02LOAM42.1736.7839.1210.24
      LeGO-LOAM39.2337.2738.1212.23
      FAST-LIO6.235.245.462.52
      Proposed6.524.395.212.81
      10LOAM3.794.984.654.25
      LeGO-LOAM4.024.395.103.16
      FAST-LIO3.763.984.142.57
      Proposed1.481.791.290.69
    • Table 2. Ablation experiment of ground segmentation and loop optimization removal module based on KITTI dataset

      View table

      Table 2. Ablation experiment of ground segmentation and loop optimization removal module based on KITTI dataset

      SequenceLoopRMSE /m
      LeGO-LOAMLeGO-LOAM*Rate of improvement /%Radar odometerProposedRate of improvement /%
      0019.155.1673.0518.433.2782.25
      0239.237.7980.1437.916.5282.80
      085.043.1437.694.592.6741.83
      0924.908.1167.4222.986.6271.19
      01129.9587.7932.44127.5776.9739.66
      104.022.8728.603.562.1240.44
    • Table 3. Comparison of average single-frame point cloud data processing performance under different algorithms

      View table

      Table 3. Comparison of average single-frame point cloud data processing performance under different algorithms

      AlgorithmProcessing frame (number/frame)Segmentation (ms/frame)Time (ms/frame)Time (ms/loop frame)
      LOAM35678100
      LeGO-LOAM347322564180
      FAST-LIO87829110
      Proposed4582126.765.5150.2
    Tools

    Get Citation

    Copy Citation Text

    Zhaoqiang Li, Yue Zhang, Fuli Xiong, Huijie Su. Laser Simultaneous Localization and Mapping Algorithm Based on Optimized Ground Segmentation and Closed-Loop Detection[J]. Acta Optica Sinica, 2024, 44(20): 2015001

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category: Machine Vision

    Received: Apr. 8, 2024

    Accepted: May. 27, 2024

    Published Online: Oct. 11, 2024

    The Author Email: Zhang Yue (1825672157@qq.com)

    DOI:10.3788/AOS240817

    CSTR:32393.14.AOS240817

    Topics