GNSS-based unscented filtering for road map databases management

This work highlights a method to manage road map databases in real-time. This method is able to detect missing roads and to update the databases from GPS data. It relies on an unscented Kalman filter whose main benefit is to deal with non-linear equations directly and to fuse available GPS and road map data in a centralized fusion scheme. We modeled the digital road map as an additional sensor which allows to account for the errors and uncertainties of the database. A probabilistic map-matching algorithm 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 We show experimental results from an urban transport network scenario where roads are missing due to an outdated map database.