For autonomous vehicles on the road, one of their primary tasks is to position the vehicle on the road. To do this, the vehicle needs to consider information from multiple sensors and fuse this information with data from road maps. The road localization problem can be broken down into three parts: The first part is to determine the road on which the vehicle is currently traveling. In fact, the Global Navigation Satellite System is not accurate enough to infer this information on its own, so a filtering step is needed. The second part is to estimate the vehicle’s position in the lane. The third part is to evaluate the lane in which the vehicle is currently traveling. The discussion here is mainly limited to the last two parts, as the first part already has mature solutions in localization based on navigation maps, such as map matching (matching the vehicle’s position with the road network on the map), etc. This chapter is a continuation of Chap.  6 , first introducing low-cost lane-line map-based localization technology, then discussing laser radar-based localization algorithms, followed by an analysis of multi-sensor fusion localization methods, and finally introducing deep learning-based localization methods.