GPS Navigation Methods Research Based On High Precision Lane-level Map

The widely use of electronic map in the in-car navigation system has promoted the development of high precision navigation map data in many countries. But nowadays most of the electronic map data production methods can only acquire that of road level. With the development of Intelligent transportation systems(ITS), advanced driving assistance systems(ADAS) and intelligent vehicle technologies, more detailed map data with higher precision is required. Recently many studies need to use high precision electronic map, such as intelligent vehicle using a priori map information for accurate positioning and lane-level path planning.High Resolution map refers to map with high precision and defined in details. The precision needs to be at least decimeter level to distinguish different lanes on the road. Nowadays with the development of positioning technology, high-precision positioning has become possible. The detailed definition means it need to contain all kinds of information in the traffic scene, including the road network data in traditional maps, lane network data and traffic signs. The main works of this paper are as follows.First of all, an overall framework of map data extraction and production method based on panorama camera is building according to the need of autonomous cars, including the data acquisition hardware framework and the software framework. The data acquisition sensors include global positioning system(GPS),inertial navigation system(INS)and wheel speed encoder used for data fusion and the map information is extracted from the panoramic images. The software framework includes the geographic information system(GIS) and database.Next the specific methods used in the map generation process are introduced in detail, including the definition of HD-map, the accurate positioning method and panoramic image stitching etc. The concept of HD-map is put forward based on the need of intelligent vehicle navigation.At last in this paper the lane level path planning method based on the generated HD-map is discussed. The goal is to generate a passable, lane-level and optimal path using the map information.