It is essential to estimate the vehicle localization for an autonomous safety driving. In particular, since LIDAR provides precise scan data, many studies carried out to estimate the vehicle localization using LIDAR and pregenerated map. The road marking always exists on the road because of provides driving information. Therefore, it is often uses for map information. Generally, road marking is stored as the LIDAR intensity data in the grid map. However, in the case of grid map, the size of the grid should be small for a precise localization. As a result, map size is increases exponentially. In this paper, we propose to generate the Gaussian mixture map based on road-marking information and localization method using this map. Generally, the NDT scan matching method based on single Gaussian distribution is used to LIDAR sensor matching. The NDT scan matching provides good performance and fast processing time. However, since it is based a single Gaussian distribution, it cannot express complex shapes. Therefore, it is difficult to apply the road marking. On the other hand, Gaussian mixture distribution can be effectively expressed the road marking because it can be included several probability distributions in the single grid. In addition, the NDT scan matching has a problem in that the convergence time is fast when the resolution is large but the performance is also deteriorated. Conversely, when the resolution is small, the matching performance is good but the convergence time is slow and it is easy to fall into local minimum. On the other hand, since the Gaussian mixture map is represented by multi resolution map, the problem of the NDT scan matching can be solved. In this paper, we perform vehicle localization using Gaussian mixture map and analyze localization performance through the experimental result.

Keywords: autonomous vehicle, LIDAR, Gaussian mixture map, vehicle localization