关键词: 高精度地图, 隐马尔可夫模型, 自动驾驶, OpenDrive

Abstract: With the widespread production and application of high-definition maps,the construction of map model topology has always been a key step in the automated map production.To address the challenge of automatically establishing topological relationships for high-definition map in complex scenarios such as road intersections,this paper proposes a method for automatic construction of topological relationships in high-definition maps based on a hidden Markov model road network matching strategy.By combining the matched standard road navigation map to provide topological information,lane attachments are made for road intersections in various complex scenarios.Based on the map data of several cities in Ningxia and Zhejiang,analysis and verification show that the recall rate and accuracy of various intersection models meet the requirements for automated production.Experiments prove that the algorithm in this paper can efficiently and accurately calculate the topological relationships of road intersection models,while effectively reduce the cost and cycle time of mass production of high-precision maps.

Key words: high-definition map, hidden Markov model, autonomous driving, OpenDrive