Urban road localization by using multiple layer matching and high resolution 3D point cloud

Abstract

In recent years, automated vehicle researches move on to the next stage, that is, auto-driving experiments on public roads. To realize practical performance, some static information should be kept on memory such as road topology, building shape, white line and so on. Some measurement companies have already begun to provide highly-precise 3-D point cloud. This study focuses on how to realize accurate localization based on the use of LIDAR data with precise map. In different roads such as urban road and expressway, the observation information of surrounding roads is significantly different. For example, on the urban road, many buildings can be observed at the upper part of the vehicle. Such observation enables to realize accurate map matching. On the other hand, in the expressway, the upper part of the vehicle has no specific observations. Therefore, it is necessary to observe the lower part of such area for the map matching. To adapt to the situation changes, we propose a localization method which is based on multi-layered scan matching. The main idea is to effectively use the features observed in different heights. Localization experiments show that the proposed method is able to estimate accurate vehicle position in different driving areas.