Automated bias-correction for accurate FCD processing systems

Abstract

Within the European research project SimpleFleet, a fleet management system for small and medium enterprises has been developed. GPS traces are collected from equipped vehicles and fed into an FCD processing system to generate traffic data. To estimate the quality of derived traffic data and determine the systematic bias of a FCD processing system a previous method called “self-evaluation” was applied for specific times of day and roads with similar characteristics. With the assumption that the relative bias is the same for corresponding traffic situations, it can be transferred and used to correct results of further computations. Notice that this opens up a convenient road for automated online corrections based on really large amounts of data, rather than on small data sets from costly measurement campaigns.