A smart map representation for autonomous vehicle navigation

Autonomous vehicle technology has attracted many attention in recent years. However, this technology is still challenging in urban areas because of the complicated environments. A highly detailed map for autonomous vehicle is proposed in this paper. This map representation consists of three basic traffic elements: roads, lanes and lane markings. It can represent both the geometrical and topological information of the traffic environment. A highly detailed map database containing these three elements is generated using ArcGIS based on a LIDAR high precision map. Then, a two-class path planning including road class planning and lane class is executed on this map. The map representation proposed in this paper provides an efficient way for intelligent vehicle to pass an intersection in a safe and reasonable way.