Multi-Sensor Fusion of GNSS receivers, inertial sensor and cameras for Precise and Reliable Positioning

A precise and reliable position and attitude information is needed with high availability in many applications. In this paper, we describe a multi-sensor fusion of two GNSS receivers, a Virtual reference station, an inertial sensor, a camera and geo-referenced satellite/ aerial images. The camera and georeferenced satellite/ aerial images require a pre-processing to obtain an absolute position information. We perform a homogenous projection, apply a color transformation, a morphological operation and an edge and corner detection to extract distinctive image points, that are matched between the camera images and geo-referenced satellite/ aerial images. It is shown that the vision based position information substantially simplifies the GNSS carrier phase ambiguity resolution. The multi-sensor fusion was also verified in a test drive. We observed a substantial improvement of the positioning accuracy compared to a GPS/ INS-only solution.