Data fusion for georeferencing a laser scanner based multi-sensor system in a city environment

Abstract

Urban environments cause difficulties for direct georeferencing approaches based on GNSS. High buildings and other obstacles produce shadowing or multipath effects degrading the positioning quality or even preventing the positioning altogether. But especially in urban environments precise positioning is important when maneuvering in narrow streets with other cars and pedestrians. We present an approach to fuse the information for classical direct georeferencing approaches used for multi-sensor systems (MSS) with information gained by a data-driven georeferencing approach. This approach assigns the measurements of a laser scanner to a 3D city model and a digital terrain model to improve the pose estimation of the MSS by GPS and IMU measurements. A real dataset recorded by a carmounted MSS is used for the evaluation. The resulting trajectory is validated by comparing to a reference solution.