Laser & Optoelectronics Progress, Volume. 59, Issue 4, 0428002(2022)

Building Method of Semantic Map Based on Improved PFPN

Lijun Ren1, Yuansheng Liu2、*, and Kedi Zhong1
Author Affiliations
  • 1Beijing Key Laboratory of Information Service Engineering, Beijing Union University, Beijing 100101, China
  • 2Beijing Engineering Research Center of Smart Mechanical Innovation Design Service, Beijing 100101, China
  • show less
    Figures & Tables(16)
    Block diagram of semantic map construction
    Framework of PFPN semantic segmentation network
    Structure of reduced PFPN
    Schematic diagram of coordinate system
    Time matching method of Lidar and camera
    Fusion result
    Construction of semantic map
    Experimental test route and test environment. (a) Test route; (b) test environment
    Map comparison. (a) SLAM map established by original point cloud; (b) semantic map after removing vehicle information in environment; (c) (d) enlarged display of rectangular area in Fig. 9 (a),(b)
    Mapping track
    Error comparison between x-axis and y-axis
    Comparison of overall pose error
    • Table 1. Comparison of different methods

      View table

      Table 1. Comparison of different methods

      MethodBackbonemIoU /%Time /s
      DANetResNet-10139.7
      HRNetHRNet42.73.89
      PSPNetResNet-5039.90.36
      PFPNResNet-101-3×42.10.28
      PFPNResNet-50-3×40.20.21
      PFPNResNet-50-1×41.20.18
      Proposed methodResNet-50-1×41.00.13
    • Table 2. Average number of point clouds per frame

      View table

      Table 2. Average number of point clouds per frame

      MethodNnumber of point clouds
      Original method2089
      Proposed method1258
      Change831
    • Table 3. Comparison of average single registration time

      View table

      Table 3. Comparison of average single registration time

      MethodTime /ms
      Original method53
      Proposed method31
      Change22
    • Table 4. Average error analysis of key parameters

      View table

      Table 4. Average error analysis of key parameters

      ParameterOriginal methodProposed methodChange
      x /m0.43880.32980.1090
      y /m0.35080.23710.1137
      P /rad0.24580.16140.0844
    Tools

    Get Citation

    Copy Citation Text

    Lijun Ren, Yuansheng Liu, Kedi Zhong. Building Method of Semantic Map Based on Improved PFPN[J]. Laser & Optoelectronics Progress, 2022, 59(4): 0428002

    Download Citation

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

    Category: Remote Sensing and Sensors

    Received: Mar. 8, 2021

    Accepted: Apr. 2, 2021

    Published Online: Feb. 15, 2022

    The Author Email: Yuansheng Liu (yuansheng@buu.edu.cn)

    DOI:10.3788/LOP202259.0428002

    Topics