High-precision positioning is the basis of intelligent vehicle-road coordination. The accuracy of the traditional positioning system is insufficient for intelligent vehicle positioning in scenarios such as vehicle interaction. Differential GPS locator and inertial navigation significantly improve the positioning accuracy, but the cost is much higher. In this paper, an aided positioning method using machine vision to identify road markers through on-board camera is proposed, which can improve the accuracy of satellite positioning by using intelligent terminals to obtain the vector changes of road landscape markers and vehicles. Firstly, based on the vehicle longitudinal dynamics model, establish the dynamic geometric vector model of vehicle and road marking points. Then, obtain the correction amount of vehicle lateral migration by using the built-in gyroscope of the intelligent terminal. Finally, combined with the regional landscape markers covered by satellite positioning, the current accurate position of the vehicle is derived backward. In the simulation process, the relative position of vehicle and road markers in the process of vehicle driving is simulated. After the traditional positioning system locates the vehicle, the method of intelligent identification is adopted to correct, and the positioning result after algorithm correction is obtained. The test results of simulation show that the error accumulated in this positioning method is slight, and the error measured can be reduced to submeter level. On the basis of continuously improving the mark points of roadbed, this method can be applied in scenarios such as vehicle interaction, and it can also be widely used in the intelligent speed planning.
Vehicle Aided Positioning Method Based on Intelligent Identification
2019-06-01
687893 byte
Conference paper
Electronic Resource
English
Intelligent vehicle positioning identification method and device based on deep learning
European Patent Office | 2023
|