GNSS-BASED RELATIVE LOCALIZATION FOR URBAN TRANSPORT APPLICATIONS WITHIN THE COVEL PROJECT

摘要

Reliable and accurate localization systems are a crucial requirement for modern Intelligent Transport Systems applications. While advanced systems which are based on dual-frequency GNSS-receivers with differential support are available, no sufficient low-cost solutions exist up to now. The CoVeL project aims to close this gap by introducing a sophisticated positioning algorithm which utilizes single-frequency off-the-shelf GPS-receivers in addition with corrections from EGNOS/EDAS and advanced map matching. The scope of this paper is the description of the Relative Positioning component of the CoVeL system. It will be shown, how exchanged GNSS raw data-so called pseudoranges -can be used for generating an accurate relative vector between participating vehicles. The system uses a Bayesian filter implementation to estimate the relative vehicle position. The performance of the proposed algorithm will be compared to a naive subtraction of pre-calculated positions. For this evaluation real sensor data and high-accurate ground truth data from a GPS/INS system were used.