An Innovative Hybrid Model for Analysis of Highway Sight Distance Using LiDAR Data

High-resolution LiDAR data, which provide a rather close fit to the real 3D environment, have been used to estimate available sight distance (ASD) along existing highways on computers in place of risky field measurements. However, the computational efficiency of existing models of ASD assessment is still unsatisfactory owing to the extremely large volume of LiDAR data as well as the burdensome manner of visibility examination implemented in previous studies.To tackle this problem, an innovative hybrid model, which is based on a newly-proposed algorithm of visibility analysis and a BP neural network, was developed to analyze highway ASD intelligently and efficiently using LiDAR data. The visibility-checking algorithm that combines cylindrical perspective projection and modified Delaunay triangulation not only enables collection of numerous labeled training data for the neural network, but also accounts for generating successive vision images along with the ASD estimation while the neural network trained using 416721 collected data is used to improve the efficiency of assessment.The results imply that the hybrid model can serve as an effective tool for highway ASD evaluation using LiDAR data. Compared with existing models, the computational efficiency of the new model is substantially improved, which is beneficial to its practical application in large-scale projects. Additionally, dynamic visualization of the estimation process enabled by the hybrid model can aid in real-time monitoring of the calculating results of ASD, providing a better understanding of visual conditions along the given highway.

  • Supplemental Notes:
    • This paper was sponsored by TRB committee AFB80 Standing Committee on Geospatial Data Acquisition Technologies. Alternate title: An Innovative Hybrid Model for Intelligent Analysis of Highway Sight Distance Using LiDAR Data.
  • Corporate Authors:

    Transportation Research Board

    ,    
  • Authors:
    • Ma, Yang
    • Zheng, Yubing
    • Cheng, Jianchuan
    • Zhang, Yunlong
    • Han, Wenquan
  • Conference:
  • Date: 2019

Language

  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures;
  • Pagination: 8p

Subject/Index Terms

Filing Info

  • Accession Number: 01697983
  • Record Type: Publication
  • Report/Paper Numbers: 19-01572
  • Files: TRIS, TRB, ATRI
  • Created Date: Dec 7 2018 9:43AM