Model-based navigation using GPS: One step closer to intelligent, incremental maps

We present the algorithmic concepts and the experiences with our prototype implementation of a software system to create navigation map data out of individually recorded GPS tracks. After motivating the need for personalized and incrementally extensible maps, we describe how to create routes from (multiply recorded) tracks and how to combine such routes into a navigable map graph. We focus on the models involved and the algorithms needed for creating the most probable map scenario out of the given set of basic data. We conclude with first experiences from our prototype implementation.