Exploring vehicle probe data as a resource to enhance network-wide traffic volume estimates

Abstract

Annual average daily traffic (AADT) is a fundamental input for numerous civil engineering applications, yet generating reliable estimates of AADT at a network-wide level poses challenges. This article explores the potential use of vehicle probe data to enhance conventional traffic monitoring practice for generating network-wide estimates of AADT by exploring relationships between site-specific traffic volume data and vehicle probe data collected in Manitoba, Canada.
 The analysis revealed that mean travel speed cannot be used to predict traffic volumes on Manitoba highways, since the mean travel speed did not deviate from the free-flow speed regardless of the volume measured. The quantity of probe data observations showed moderate correlation with traffic volume at some sites (R-squared up to 0.65), but these correlations were stronger (R-squared up to 0.9) when considering trucks only. These findings suggest that probe data could be used to estimate truck volumes at certain locations.

Statewide Truck Volume Estimation Using Probe Vehicle Data and Machine Learning

“…Zhang and Chen ( 14 ) found a very strong correlation between overall traffic volumes and probe vehicle counts. Grande et al ( 18 ) found that truck volumes exhibited a similarly strong relationship. No previous study has accounted for this important feature.…”