High-precision map services are an indispensable basis of many innovative in-vehicle applications. Safety and comfort can be improved even further by inputting contextual information, such as road surface. Correct allocation requires lane-accurate vehicle localization. This paper presents a novel algorithm for lane-level map-matching that integrates a global navigation satellite system, an inertial measurement unit, and map data. The implemented algorithm consists of five consecutive modules: map generation, sensor-data pre-processing, road assignment, maneuver recognition, and information fusion. The approach was tested over a total distance of 245 km, involving 237 lane changes, within the metropolitan area of Munich. A total accuracy of 82% correctly classified lanes was achieved in test drives recorded with smartphones. Both the implementation and part of the data set of this paper are publicly available (https://github.com/TUMFTM/Lane_Level_Matching).     «

High-precision map services are an indispensable basis of many innovative in-vehicle applications. Safety and comfort can be improved even further by inputting contextual information, such as road surface. Correct allocation requires lane-accurate vehicle localization. This paper presents a novel algorithm for lane-level map-matching that integrates a global navigation satellite system, an inertial measurement unit, and map data. The implemented algorithm consists of five consecutive modules: ma...     »