Lane-Level Positioning using Particle Filtering
Résumé
GPS positioning and map-matching are key technologies used in Car Navigation Systems. Next generation of Cooperative Vehicle Infrastructure Systems (CVIS) will surely require positioning component with higher accuracy, integrity and availability, for instance for the needs of lanelevel applications. The concept of Map Aided Location (MAL) was introduced recently: it consists in considering the map contents (geometry and topology) as constraints in the positioning problem. The main contribution of the work reported in this article concerns the evolution model of the vehicle : whereas previous research investigations were based on usual models in plane coordinates, we propose here an evolution model adapted to the local curvature of the road. Therefore, road constraints are emphasized, particularly when GPS outages occur. An enhanced map (Emap) - where all lanes are represented with decimeter accuracy - is used in this Map Aided Location process.