抄録
This paper describes a Non-Line-of-Sight (NLOS) satellites detection method for improving GNSS (Global Navigation Satellite System) positioning accuracy in urban area. In general, NLOS satellites which are shielded by obstacle such as buildings and elevated bridges cause large multipath errors in positioning results. Therefore, our method extracts obstacle area in a fish-eye camera image by projecting 3D map on it, and detects NLOS satellites existed on obstacle area. However, camera's position and orientation on 3D map coordinate are essential in projecting 3D map, and we estimate them using particle filter. As a result of evaluation test in urban area, it is confirmed the proposed method is effective on improving positioning accuracy.
