Optics and Precision Engineering, Volume. 33, Issue 6, 993(2025)

Pose decoupled RGBD-SLAM based on point-line-plane features

Gang YANG, Wengang ZHANG*, and Tianle CAO
Author Affiliations
  • College of Communication and Information Engineering, Xi'an University of Posts and Telecommunications, Xi'an710121, China
  • show less
    Figures & Tables(18)
    Overall framework of the system
    Illustration of line segment merging
    Plucker coordinates of line
    Reprojection error model for line features
    MF and MW Scene
    Estimation of rotation based on the MF
    Values of APE RMSE for different parameters
    Line feature extraction effect of each algorithm
    Iteration plot of the APE RMSE
    Trajectory plots of each SLAM algorithm on partial sequences
    Plot of ATE between SP-SLAM and Ours on lr-kt1
    Map construction on the fr3-long-office sequence
    Map construction on the lr-kt2 sequence
    • Table 1. Each algorithm line feature extraction data

      View table
      View in Article

      Table 1. Each algorithm line feature extraction data

      图像

      序列

      LSD原EDLines改进EDlines
      时间数量时间数量时间数量
      lr-kt036.620312.226713.631
      of-kt035.223511.130812.343
      fr1-room41.754915.566717.253
      fr3-long-office41.151814.964816.656
      均值38.7376.313.4472.514.945.8
    • Table 2. Effect of line feature quality on trajectory accuracy

      View table
      View in Article

      Table 2. Effect of line feature quality on trajectory accuracy

      图像序列LSD原EDLines改进EDlines
      of-kt04.154.333.70
      fr3-long-office1.871.981.17
    • Table 3. Trajectory error comparison of different SLAM algorithms in the ICL-NUIM dataset

      View table
      View in Article

      Table 3. Trajectory error comparison of different SLAM algorithms in the ICL-NUIM dataset

      图像序列ORB-SLAM2PL-SLAMSP-SLAMOurs
      ATERTEATERTEATERTEATERTE
      客厅场景lr-kt00.750.400.660.600.850.580.650.33
      lr-kt112.071.37--6.901.132.760.51
      lr-kt21.340.453.540.891.910.501.200.41
      lr-kt31.240.480.950.622.480.471.490.44
      办公室场景of-kt02.700.542.081.072.300.503.700.40
      of-kt17.831.37--5.410.885.010.55
      of-kt21.210.592.150.681.020.531.340.43
      of-kt312.431.318.264.124.210.951.820.61
      均值4.950.812.941.333.140.692.250.46
    • Table 4. Trajectory error comparison of different SLAM algorithms in the TUM RGB-D dataset

      View table
      View in Article

      Table 4. Trajectory error comparison of different SLAM algorithms in the TUM RGB-D dataset

      图像序列ORB-SLAM2PL-SLAMSP-SLAMOurs
      ATERTEATERTEATERTEATERTE

      纹理丰富的

      非结构化场景

      fr1-xyz0.980.591.231.910.980.600.950.51
      fr1-desk1.611.011.141.431.790.891.760.63
      fr2-xyz0.390.210.430.120.370.210.460.30
      fr2-desk1.110.660.921.152.730.611.970.72
      fr3-long-office2.000.522.270.461.450.491.170.41

      纹理丰富的

      结构化场景

      fr3-str-tex-far1.041.200.910.560.971.120.900.53
      fr3-str-tex-near1.470.881.480.891.770.840.960.55

      纹理缺失的

      结构化场景

      fr3-str-ntex-far----1.781.170.850.46
      fr3-str-ntex-near----1.711.181.220.95
      动态场景fr3_sitting_halfsphere1.490.841.310.951.260.871.150.77
      fr3_walking_halfsphere53.95.3347.64.8643.72.6740.32.41
      均值7.111.256.371.375.320.974.700.75
    • Table 5. Running time of the SLAM system

      View table
      View in Article

      Table 5. Running time of the SLAM system

      程序流程主要部分SP-SLAMOurs
      跟踪点线面特征提取及匹配/ms9.419.5
      位姿估计/ms20.334.2
      总计/ms29.753.7
      局部建图局部建图及BA优化/ms125.1213.4
      回环全局优化/ms291.4-
    Tools

    Get Citation

    Copy Citation Text

    Gang YANG, Wengang ZHANG, Tianle CAO. Pose decoupled RGBD-SLAM based on point-line-plane features[J]. Optics and Precision Engineering, 2025, 33(6): 993

    Download Citation

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

    Category:

    Received: Aug. 6, 2024

    Accepted: --

    Published Online: Jun. 16, 2025

    The Author Email: Wengang ZHANG (wengz0208@163.com)

    DOI:10.37188/OPE.20253306.0993

    Topics