Estimation of truck origin-destination flows using GPS data

Large trucking vehicles have a comparatively more significant impact on safety, traffic congestion, pollution, and pavement wear than passenger vehicles. Appropriate planning and operation of truck movement are necessary to reduce these impacts. While heavy truck movement has traditionally been measured through surveys, these remain limited because they are costly and time-consuming. In this study, we propose the use of large streams of GPS data to estimate truck origin-destination flows. Large streams of GPS data have typically been difficult to use as they lack descriptors for key events during a trip unless the data is accompanied by travel diaries. We address this problem by developing a heuristic-based approach to identify the key events, such as truck stops, trips, and other trucking activities. Then, a Pearson correlation coefficient and an entropy measure are applied to compare trucks' mobility patterns and to determine whether changes in trucks travel patterns have occurred over one year. Finally, we use a multinomial logit structure to estimate destination choice models for five time periods. This research provides a strong case study of how GPS data can be used along with outputs of existing travel demand model (a model created with data collected using traditional techniques) to estimate origin-destination and destination choice models of truck movement in a provincial model setting.

Automated summary

This study proposes the use of GPS data to estimate truck origin-destination flows, develops a heuristic-based approach to identify key events, compares trucks' mobility patterns using correlation coefficient and entropy measure, and estimates destination choice models using a multinomial logit structure.