Position determination based on sensor data fusion with digital maps and 3D point cloud data

Future navigation systems will increasingly affect the operation of semi and highly automated driving. This requires a significant increase in performance and accuracy of vehicle positioning. Traditional GNSS based solutions are too inaccurate for the requirements of automated driving and are subject to significant urban deviations. This paper presents a method that supports current positioning by a LiDAR sensor. The processed point data are referenced to conventional and high-precision maps. By filtering the map data, landmarks are extracted in the form of building edges and infrastructure objects, which are used to improve the current vehicle position. The detection and tracking of the multimodal target objects is implemented via a particle filter which facilitates the determination of the absolute position by using the relative distances to a minimum of two detected reference points.