Travel Time Estimation on Road Networks

• Activity Number: 574
• Type: Topic Contributed
• Date/Time: Wednesday, August 4, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Graphics
• Author(s): Bradford Westgate*+
• Companies: Cornell University

Abstract

Estimation of ambulance travel times on road networks is critical in Emergency Medical Services planning. In addition to being used to choose the fastest route for an ambulance to answer a call, travel times are used for optimizing ambulance base locations. Unfortunately, while the optimization algorithms are quite sophisticated, the methods used to estimate the travel time inputs to those algorithms typically are not. We use a voluminous untapped data source, GPS recordings during ambulance trips, to improve estimation of travel times. Interest lies in estimating the mean and variance of the travel time on each road segment. However, the actual routes and travel times for each ambulance trip are not known exactly due to noise and sparseness of the GPS readings. We use Bayesian methods to simultaneously estimate the unknown route and road segment parameters.