Abstract: The high-definition map can provide accurate and reliable global perception and positioning information for autonomous vehicles. Therefore, it is of great significance to vehicle decision planning. However, the proposed high-definition map models generally lack reasonable and effective geometric constraints and expressions on the connected areas of intersections. Aiming at the turning driving task of vehicles at the intersection, on the one hand, based on the safe driving rules at the intersection, on the other hand, to ensure the safe passage of vehicles at the intersection, this paper proposes a method for generating high-definition maps of safe driving areas at intersections. This method aims to make up the information gap in the intersection area of the current high-definition map in the urban structured road scene. The generated safe-driving area can serve for vehicle decision planning of different automation levels and improve the safety and standardization of vehicles passing at intersections. Finally, this paper uses the real intersection data of Wuhan city to verify the proposed method. The experimental results show that in 90% of cases, the proposed method can generate the intersection safe-driving areas at experimental intersections. It proves the rationality and validity of the method in this paper.

Key words: high-definition map, intersection, regional constraints, security, normative