3D environment modeling for GPS multipath detection in urban areas

Abstract

Accurate and reliable vehicle localization with GPS is an important part of many automotive applications. Unfortunately, in urban areas multipath is a major source for positioning errors. While this can be handled with expensive equipment, vehicles sensor budget is usually limited, so other approaches are to be preferred as well. This paper aims to evaluate a multipath mitigation technique based on cheap GPS hardware combined with a 3D environment model. This model shall map surrounding obstacles like buildings and the terrain that impair the received GPS signals. As a result, this model can be used within a Bayesian implementation of a localization filter to improve the positioning accuracy and integrity.