Many Advanced Driver Assistance Systems rely on an accurate self-localization. This is usually provided by Global Navigation Satellite Systems (GNSS) along with on-board sensors, such as inertial sensors, speed sensors, wheel-tick sensors and steering wheel angle sensors. However, this approach suffers from incremental error growth when long GNSS outages occur. Additionally, it is widely accepted that GNSS has a poor performance in urban-like environments due to satellite line-of-sight blockage, signal attenuation and multipath propagation. We propose a solution in which the error associated to GNSS-based positioning is contained by using surrounding road infrastructure objects (RIO) that are detected with a radar sensor. Since the position of these objects is a-priori not known, we suggest to share their estimated location among the vehicles using vehicle-to-vehicle communication and, in this way, improve their over-all position accuracy over time. In this way, vehicles entering a GNSS-denied area are able to maintain their position accuracy, achieving even better results as if GNSS was available in the area of interest.