Incremental map generation with GPS traces

Abstract

This paper provides a generic approach for map generation based on Global Positioning System (GPS) traces especially for unknown terrains. The system uses AI algorithms to infer the road geometry. Given a set of GPS traces, this input is refined by filtering and partitioning algorithms. For constructing the graph, clustering algorithms are applied that allow to incrementally generate a road map, annotated with travel time information. The processing of the input data relies on a subdivision of the world into tiles. A flexible map viewer is provided to navigate through the hierarchically organized content of the database. The system provides the export of maps to serve e.g. as the input for a routing server, which in turn can be queried via TCP/IP for shortest and quickest routes.