Travel time estimation in Urban networks using buses as probes

Abstract

Using transit vehicles as probes offers a number of advantages as they cover a large portion of urban networks and the equipment required for data collection is usually already installed by transit operators. Despite the fact that transit vehicles and automobiles have different running behaviours, a relationship can be developed to estimate auto travel times using transit data. Travel time estimation using buses as probes is usually limited to their travel routes. This research investigates the potential of using bus travel time data to estimate general link travel times of neighbour (nearby) links. The main research hypothesis is that travel times of nearby links have strong correlation as these links are subject to similar traffic conditions. A general methodology is presented for travel time estimation using historical travel time data of the link itself and real-time bus data from neighbour links. A case study was undertaken using a VISSIM microsimulation model of downtown Vancouver. The model was calibrated and validated using real-life traffic volumes and travel time data. Travel time estimation accuracy was assessed using the Mean Absolute Percentage Error (MAPE), the value of which was 17.6%. The method was proven to be useful to estimate travel time on links that do not have real-time travel time data while having strong travel time correlation with neighbouring links.