3D Modeling of Urban Environments for Enhancing GPS Localization's Accuracy

Described project aims at accessing accurate localization. It consists in improving accuracy of localization systems with video perception of the environment around equipped vehicles. The environment corresponds to the propagation area of GNSS receiver installed on top of the vehicle. Accuracy problems are linked to obstacles density, especially in urban environments. The objective is to reduce multipath effect on the measurements, without lowering the availability of the service, which is typically the case for errors detection procedures. The presented work aims at using two fisheye cameras in a stereoscopic configuration in order to compute 3D coordinates of the scene’s points. A reconstructed 3D model allows then to correct the propagation error thanks to the knowledge of the satellites’ reception status.