Robust Vehicle Localization and Integrity Monitoring Based on Spatial Feature Constrained PF

摘要

The integrity monitoring of positioning solutions provided by the Global Navigation Satellite System (GNSS) is becoming more important, mostly due to the rapid development of autonomous vehicles and Intelligent Transport Systems (ITS) that heavily rely on GNSS positioning. Since GNSS has performance issues in urban environments where the number of visible satellites may be limited, integrity monitoring algorithms have been combined with other sensors and map matching (MM) algorithms to adjust the positioning solution when GNSS does not perform well. Integration with MM is based on the assumption that the driving vehicle is always going to be positioned on the road. However, most of these approaches use locally collected road data which limits their applicability in other locations and may give biased results due to unrealistically accurate road data. This study combines the Bayesian Receiver Autonomous Integrity Method (BRAIM) with the globally available map data set OpenStreetMap (OSM) for MM. The OSM is used as a source of road centrelines and attribute information from which the road polygons are constructed. The road polygons are used to constrain the Particle Filter (PF) particles to the surface of the road. This approach has been tested in three different driving environments in Melbourne: Highway, open-sky, and urban canyon. The best median achieved integrity in this study was 2.6·10⁻⁴ for required horizontal alarm limit of 5 m, for MM and BRAIM integration. This indicates that in conditions of good satellite visibility and measurement quality, this system could be used for payment-critical applications. However, some challenges still remain in areas where map data is incomplete or incorrect.