The Research Of Vehicle Positioning Based On Adaptive Cubature Kalman Filter Algorithm

With the continuous development of modern society, people’s demand for Intelligent Transportation System(ITS) is also rising. Vehicle positioning information is a foundation of ITS to offer many services, so it is necessary to accurately acquire the position of the vehicle. In order to improve the positioning accuracy of the vehicle, it can use advanced filtering techniques to process the measuring data to improve the accuracy.In this thesis, it will establish a nonlinear system model and use the nonlinear Cubature Kalman Filter(CKF) algorithm to estimate the position of the vehicle by the data obtained from GPS and odometer. Through the simulation, it can be found that the using of the CKF can better locate the vehicle to some degree. However, the positioning error will suddenly increase when the mutation occurs in the vehicle motion. To solve this problem, an adaptive filtering algorithm called Imm-CKF which combines the multiple models algorithm with the CKF is proposed. Through the simulation, it can be found that the using of the Imm-CKF is capable of tracking the moving vehicle when the motion mutation occurs. And in the simulation considering the general vehicle movement in this thesis, the positioning accuracy using Imm-CKF could be increased by about 25% when compared with CKF, the increase will be more obvious when the mutation is more severe.This thesis also simulates a case that the vehicle moves in the tunnel without GPS signal. In this case, the system continues using the Imm-CKF whose preliminary position information of the vehicle is obtained from the GIS database and the odometer. It can be shown that the Imm-CKF can also acquire good positioning effect in the whole process of the GPS signal present, the GPS signal disappearing, and the GPS signal recovering. In addition, this thesis considers the case that the error of the vehicle positioning in the tunnel will be cumulative. From the time when the vehicle gets out of the tunnel and can receive the GPS signal, it will take about 20 filtering cycles to fix the cumulative error and correct the position of the vehicle, so that the vehicle positioning can be stable and sustainable.The research in this thesis can not only provide a reference to the study of nonlinear filtering algorithm but also the ITS of how to accurately get the position of the vehicle.