A simple data fusion method for instantaneous travel time estimation

Travel time is one of the most understandable parameters to describe traffic condition and an important input to many intelligent transportation systems applications. Direct measurement from Electronic Toll Collection (ETC) system is promising but the data arrives too late, only after the vehicles complete their trip. There are several existing models with varying degree of success to indirectly estimate travel time from loop detector. The performance of these models depends significantly on the variation of traffic condition. By closely looking at the time-series of the estimated travel time with the actual travel time, the error was found to follow a specific pattern for each traffic condition. The goal of this research is to develop a simple data fusion between loop detector data and ETC data to make more accurate estimation of instantaneous travel time on expressway corridor. With the error pattern for each traffic condition in mind, it is possible to develop a simple fusion method that can improve the accuracy of travel time estimate even under sparely distributed detectors.