Integrating Non-Motorist Facility Data into Comprehensive Road Safety Assessment

Details

• Creators:
  Qin, Xiao ; Schneider, Robert J ; Sayed, Md Abu ; Vajari, Mohammad Abrari
• Corporate Creators:
  Center for Pedestrian and Bicyclist Safety (CPBS) Tier-1 University Transportation Center (UTC) ; University of Wisconsin-Milwaukee. Institute for Physical Infrastructure and Transportation
• Corporate Contributors:
  United States. Department of Transportation. University Transportation Centers (UTC) Program ; United States. Department of Transportation. Office of the Assistant Secretary for Research and Technology
• Subject/TRT Terms:
  Aerial Photography Artificial Intelligence Equity (Justice) Nonmotorized Transportation Road Markings
• Publication/ Report Number:
  23UWM02
• DOI:
  https://doi.org/10.21949/sjgp-f537
• Resource Type:
  Tech Report
• Geographical Coverage:
  United States
• Edition:
  Final Report – June 2023 to May 2024
• Corporate Publisher:
  Center for Pedestrian and Bicyclist Safety (CPBS) Tier-1 University Transportation Center (UTC)
• Abstract:
  As active transportation gains popularity for both work and leisure trips, the need for reliable and accurate data on non-motorist activities and infrastructure becomes increasingly critical. Traditionally, detailed data on non-motorist activities and facilities has been scarce and challenging to obtain. However, advancements in technology now provide an opportunity to collect these data. This study aims to investigate recent technological advancements and methodologies for collecting non-motorist data and evaluate their ability to detect, classify, monitor changes in, and assess the condition of these facilities. A key focus is the use of vision-based large language models, specifically ChatGPT, for retrieving and evaluating non-motorist facilities from aerial images. This study tested ChatGPT's performance on a dataset containing satellite images to evaluate its ability to accurately detect and identify features such as pedestrian crosswalks. The results showed that ChatGPT can reliably assess crosswalk and sidewalk conditions, informing maintenance and improvement strategies. However, limitations were observed, including inconsistent detection of bicycle lanes without explicit visual cues and challenges in classifying crosswalks into specific types. Additionally, it struggles with reliably identifying street lighting and pedestrian signals, which are critical for comprehensive safety assessments.
• Format:
  PDF
• Funding:
  69A3552348336
• Collection(s):
  University Transportation Centers
• Main Document Checksum:
  urn:sha-512:ae877209490cc4fd64bdd775cd0f77070e8b0591a00db39c0cc4816e477d11a02c8b71f8db1c50408a8c115b48f1689e4678ce4d784671796eb87adbbed4446f
• Download URL:
  https://rosap.ntl.bts.gov/view/dot/86189/dot_86189_DS1.pdf
• File Type:
  [PDF - 1.64 MB ]