A Study On Road Networks Matching For Highly Automated Driving Map And Traditional Navigation Map

Highly Automated Driving map(HAD map)plays an important role in autonomous driving systems.On the one hand,HAD map could present the entities that do not exist in traditional navigation map.On the other hand,there is richer traffic information in traditional navigation map.HAD map and traditional navigation map complement and interact with each other.Instead of using them independently,autonomous driving applications always utilize HAD map and traditional navigation map simultaneously.The mapping relationship between HAD map and traditional navigation map could enhance the complementarity and interactivity of them.Besides,the mapping relationship could also make it more flexible for autonomous driving applications to use the map data.In this study,firstly,HAD map and traditional navigation map have been introduced,including their data contents and map models.And the difference between HAD map and traditional navigation map has been explained.Considering their great difference in topology and geometry,a topology reconstruction method was proposed to make their topologic networks as similar structures.In road networks matching process,four features:distance,direction,topologic structure and geometrical morphology similarity,were utilized in intersection matching algorithm.With the mapping relationship of intersections and the reconstructed topological networks,a forward path exploring algorithm was used to collect identical road sets.Besides,a linear referencing system was established on the network of HAD map;and using dynamic segmentation technology,mapping relationship between the road networks of HAD map and traditional navigation map was built.Finally a physical data model was proposed for the mapping relationship storage.A matching system for HAD map and traditional navigation map has been developed based on Visual Studio 2010.The proposed approach was tested with two typical highway areas of HAD map.and corresponding traditional navigation map.The experiment showed that the method could efficiently perform road networks matching for HAD map and traditional navigation map.And it has been used in map production with strong robustness and applicability.