Automatic Detection of Topological Changes for Digital Road Map Updating

This paper highlights a method to detect topological changes of digital road maps automatically. Here, the road network updating can be done in real time. It relies on collected GPS data from vehicles. These data are fused with existing road map data by an unscented Kalman filter in a centralized scheme. We modeled the map data as a sensor that allows accounting for the errors and uncertainties of land surveying. The core of our method is a probabilistic map-matching approach that is used to manage the road network database through the computation of the Mahalanobis distance. We show experimental results from an urban transport network scenario in which we deal with the case of opened roads and the recent conversion of a crossroad into a roundabout. Then, the obsolete map database is updated to ensure a more reliable route planning.