Improving localization in digital maps with grid maps

This paper presents a self-localization concept for road vehicles using only a mono camera as sensor. It employs a particle filter to estimate the global position based on a digital map of lane markings. A grid map is used to smoothen the lane marking detections. This approach is compared to an approach that does not rely on a grid map as an intermediate stage. Results obtained on real world data show improvements of the localization accuracy when a grid map is employed. Furthermore, a method to create digital maps of lane markings is presented.