Autonomous Localization Method for Railway Trains Based on Multi-Source Information Fusion

Aiming at the problems of high cost and difficult maintenance in the practical application of the traditional "vehicle-ground cooperation" train positioning technology in speciaf section railway projects. In this paper, a method for autonomous positioning of railway trains based on multi-source information fusion is proposed. First, develop a kilometer mark detection thread based on VINS Fusion, and build a lightweight optical symbol recognition system to identify the character area of the kilometer mark to obtain the position information of the electronic map. Secondly, based on the LSD line detection algorithm combined with the way of dynamically expanding the line, the coordinates of the vertex coordinates of the kilometer mark are extracted. Then, a global optimization objective function containing the position information constraints of the kilometer mark is established, and the global positioning accuracy is improved through the graph optimization method. This solves the problem that existing intelligent positioning of railway trains cannot eliminate accumulated errors in the absence of loop closure detection. Finally, the semi-physical simulation results show that the positioning error of the original positioning system can be reduced by more than 57%.