Analysis of GPS Fleet Tracking Data to Infer Commercial Vehicle Travel Patterns in Ontario
Corresponding Author: Bryce Sharman, WSP
Presented By: Bryce Sharman, WSP
Abstract
WSP|Parsons Brinckerhoff is preparing two large-scale multi-modal transportation forecasting models for the Ministry of Transportation of Ontario (MTO), both of which include tour-based models of commercial truck travel. Currently, estimation and calibration of disaggregate commercial vehicle travel models is hindered due to a lack of high-quality data describing truck travel behavior. GPS fleet tracking data are becoming more widely available and show large promise for this application as they provide travel information over extended time periods with highly accurate spatial and temporal resolution. The primary difficulty using this data source is that the only available data are the position and time history of the observed vehicles.
This paper presents a novel GPS processing to convert the raw GPS data into a travel diary of trips, stops, and tours, with an emphasis on observing the behavior of urban truck travel. Particular emphasis will be placed on a multi-step stop identification procedure, and on a novel depot identification algorithm that uses clustering techniques to identify locations with many long stops. Depot identification is particularly important as it identifies “home” locations, which must be accurately identified as they are key to tour formation.
Finally, this paper will present an overview of the observed results including summary distributions of stop dwell times, trip and tour distances and durations, number of trips per tour, tour departure times and the relationship between tour departure time and duration.