Fusion of GPS, OSM and DEM Data for Estimating Road Network Elevation

This paper presents a method to estimate the roads elevation by fusing data from GPS receivers, OSM road network and DEM terrain surface. It relies on GPS data collected from a vehicle that travels the OSM road network. Also, a digital elevation model from SRTM data is combined in order to get a discrete elevation of the road. The fusion algorithm implements an unscented Kalman filter in a centralized scheme. Here, roadmaps and DEM data are modeled as measurement equations that allows to account for their errors and uncertainties. The method highlights the advantage of a probabilistic dual-matching, based on the computation of Mahalanobis distances, that allows to identify and match GPS positioning with the OSM road network and the DEM terrain surface. Experimental results show that the proposed method leads to improve the road elevation estimation with respect to conventional approaches using DEM data only.