Lane Level Path Planning for Urban Autonomous Driving using Vector Map

In the past decade, autonomous vehicles have received remarkable attention for the efficiency and safety of road traffic. For more advanced autonomous driving systems, the importance of vector maps containing location information on various objects, such as multi-lane, centerlines of the lanes, traffic signs and signals, etc. is becoming increasingly important. However, until now, vector maps are mainly used to increase localization accuracy by using perception results or visualization of the driving environment. In this study, we propose a lane level path planning method using vector maps. The proposed lane level path planning method is designed to increase path planning efficiency and flexibility by searching the global route and lane level route to destination hierarchically. In addition, techniques for minimizing manual annotation, which has been a barrier to the use of vector maps, have also been proposed. The validity of the proposed method was verified through a test using a real lane-level road network.