Statistical modelling of digital elevation models for GNSS-based navigation

Recently, smart mobility has become a important activity in transportation systems such as public, autonomous and shared transports. These systems require reliable navigation applications that lead to precise localisation and optimised route. The GPS system may face problems such as signal degradation caused by conical effects, affecting the reliability and accuracy of the signal, or signal loss in poor visibility environments. By using other sensors, the vehicle location system can overcome these GPS problems. This work focuses on the estimation of the inclination, which will be used to optimise the route planning for the EV or HEV especially in order to control the energy consumption. This paper presents a multi-sensor fusion method, based on GNSS, INS, OSM and DEM
data fused using a non-linear particle filter, to estimate and improve the slopes of road segments. A new statistical modelling of the DEM errors related to the spatial sampling of elevation data is proposed. This method is based on the definition of a geometrical window, called Adjacent Sliding Window (ASW), which dynamically selects the elevation data in the vicinity of the road. The proposed method is evaluated in a suburban transport network. The experimental results show the benefits of the vehicle attitude and road slope estimation accuracies.