Combining GPS Data Collection with Assisted Machine Learning to Enhance Freight Vehicle and Driver Surveys: Methodology and Demonstration

This paper presents an innovative freight vehicle and driver survey that combines Global Positioning System (GPS) data collection with assisted machine learning based on driver verification. This survey is able to collect accurate and rich behavior data with reduced respondents’ burden when compared to passive GPS data collection methods and/or basic stand-alone conventional freight vehicle driver interviews. This research involves devising and implementing survey methods to collect this and are designed with a GPS, generating the principal travel data coupled with a pre-survey providing key background on the freight vehicle and driver. The GPS data are then enriched with behavioral and other data using machine learning that is assisted by a driver verification process. A major advancement of this survey method is the design of a driver verification stage, which verifies the stops made by the driver and activities performed throughout the trips. Additional contribution of this research is the enhancement of the survey for applicability across various real-world scenarios. This includes participation by combinations of carriers, stand-alone drivers, vehicles driven by multiple drivers, or drivers operating multiple vehicles. The verification methodologies has been implemented in both mobile app-based and web-based versions, and designed with temporal and geographic scalability. To demonstrate the applicability and value of this survey method, a small-scale pilot study in the Boston area is presented. The survey improvements reveal comprehensive insights into freight activity and driver behavior, which in turn permits more behavioral realism in freight transportation models and other types of analysis.

  • Supplemental Notes:
    • This paper was sponsored by TRB committee ABJ90 Standing Committee on Freight Transportation Data.
  • Authors:
    • Ding-Mastera, Jing
    • Jing, Peiyu
    • Stinson, Monique
    • Manzi, Eric
    • Zhao, Fang
    • Marzano, Vittorio
    • Cheah, Lynette
    • Ben-Akiva, Moshe
  • Conference:
  • Date: 2018


  • English

Media Info

  • Media Type: Digital/other
  • Features: Figures; Maps; References; Tables;
  • Pagination: 21p

Subject/Index Terms

Filing Info

  • Accession Number: 01661013
  • Record Type: Publication
  • Report/Paper Numbers: 18-00512
  • Files: TRIS, TRB, ATRI
  • Created Date: Feb 23 2018 4:27PM