A Precise Road Network Modeling and Map Matching for Vehicle Navigation

In this paper, we describe a precise road model for vehicle localization and map matching for the purpose of driver assistance. This model utilizes the prevalent digital road map and road points ('IP' points) such as: intersection points and supporting points with orientation variation, to estimate road curvature in two-dimensional alignment. In our work, road geometries such as clothoid curve and circle are used to reconstruct road shape, which was approximated by piece-wise lines before. Therefore, we design the whole network modeling consisting of 'IP' points, linear segments and curve segments. And new map matching approach is also proposed according to different segments. Finally, in order to test the performance of this model in practical situation, we employed VICNAS system with new hybrid vehicle localization approach combining commercial GPS and inertial sensors, and experimental results verified the effectiveness and efficiency of our approach.