3D LIDAR Based Vehicle Localization using Vertical Structure Landmark in Urban Road

Jun-Hyuck Im, Kyu-Won Kim, and Gyu-In Jee

Abstract:	Because tall buildings in urban areas obstruct the GPS satellite signal reception, the position accuracy of the GPS is very poor. However, buildings in urban areas can be used to generate a landmark map for precise vehicle localization. This map’s information can be represented in various forms. We focused on the outer wall of a building which is erected vertically to the ground and almost flat. Therefore, the vertical corners that meet the vertical planes are present everywhere in urban areas. These corners can provide very good landmarks and also can be extracted using the LIDAR. Additionally, traffic sign can be used for localization. The traffic signs are present in the road and reflect light well. Therefore, the traffic signs can be clearly distinguished from other structures by using the LIDAR reflectivity. In this paper, we used building’s vertical corner and traffic sign for vehicle localization in an urban area. Also, we proposed a vertical structure landmark map with information for vertical corner and traffic sign. The experiment was carried out in the Gangnam area of Seoul, South Korea. The traveling distance was about 4.5 km and the maximum traveling speed was about 80km/h. The lateral and longitudinal RMS position errors were 0.118m and 0.231m, respectively.
Published in:	Proceedings of the 30th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2017) September 25 - 29, 2017 Oregon Convention Center Portland, Oregon
Pages:	387 - 401
Cite this article:	Im, Jun-Hyuck, Kim, Kyu-Won, Jee, Gyu-In, "3D LIDAR Based Vehicle Localization using Vertical Structure Landmark in Urban Road," Proceedings of the 30th International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2017), Portland, Oregon, September 2017, pp. 387-401. https://doi.org/10.33012/2017.15156
Full Paper:	ION Members/Non-Members: 1 Download Credit Sign In