Satellite positioning performances are critical for transport liability applications such as road user charging, or pay per use assurance. The scientific community has focused for several years on environments with favorable conditions (aviation, maritime, rural) and is now addressing transportapplications including urban areas. This is the case of this paper that presents the result of a study co-financed by the National French Space Agency (CNES). This study aimed first to collect the positioning data in urban area of several GPS/EGNOS receivers (more than 1.5 millions of positions) in addition to their associated satellite raw data and the reference positions. Its goal was secondly to assess the performance using classical positioning algorithms and last to propose and assess improved algorithms. The authors present first how to obtain the true trajectory necessary for the performance assessments. The instrumentation they set-up on board a test vehicle is based on IXSEA LandINS inertial measurement unit hybridized with kinematics GPS and odometry. The reference trajectory obtained is analyzed, with the aim of qualifying its accuracy. Based on these measurements, the authors then present the performance improvements obtained using different techniques, in particular 2D techniques to improve availability in difficult area such as narrow streets, smoothing technique to optimize thefiltering and local error modeling technique to assess in particular disturbing multipath effects. Thanks to the ground truth obtained in this benchmark, the authors demonstrate that these techniques bring a significant improvement to position data performances, enlarging that way the spectrum of liability-critical road transport applications in urban areas.
Abstract