Application Research Of GNSS/INS Fusion Positioning Filter Method

Driverless technology and artificial intelligence have become the cutting-edge technologies of people's research nowadays.In the perception level of smart cars,the importance of positioning is self-evident.in order to accurately drive on the road Intelligent cars need to know their precise position relative to the external environment.The combination of Global Positioning System(GNSS)and Inertial Navigation System(INS)can provide high-precision navigation and positioning information,which is one of the best solutions for smart cars in positioning,and the selection of appropriate filtering algorithm is also crucial in sensor fusion.Therefore,in the development of smart car technology,it is of great significance to research improve the positioning accuracy.For the positioning of smart cars,this paper focuses on the fusion filtering method of GNSS/INS sensors and its application in Beidou smart cars.The main research work is as follows:(1)The composition of global navigation satellite system and satellite orbit model is introduced.A satellite-borne GNSS observation error model of Beidou intelligent vehicle was established.At the same time,the PVT information calculation method and coordinate transformation method of user receiver are introduced in detail.(2)Various GNSS positioning methods,including PPP,RTK and RTD,are introduced systematically.The basic principle and error sources of inertial navigation system are combed,the positioning principle of GNSS/INS combination system is expounded,the loose combination model is selected as the sensor combination mode,and the fusion positioning of different sensors is realized.At the same time,the principle of sensor fusion and the commonly used method of sensor data fusion are introduced.(3)By analyzing the filtering theory and its applicable scope of KF,EKF and UKF filtering models,EKF filtering model and UKF filtering model are constructed respectively.In addition,the relevant filtering algorithm was studied,and the experimental scene was set to simulate the road in the complex environment.After the combined filtering of GNSS/INS,the position error is within 1m,and the speed erroris maintained at about 0.5m/s.Compared with the position error before the filtering,the position error is increased by 75.10% and the speed error by 51.60%.Compared with before the filtering,the accuracy of position and speed are improved.(4)GNSS sensors and INS sensors mount on Beidou smart car are used to obtain measured data,and EKF and UKF filtering algorithms are applied to GNSS/INS fusion positioning.By comparing the two filtering results,UKF filtering results are compared with EKF filtering results,the standard deviations in the east and north directions are reduced by 0.062 m and 0.188 m respectively,horizontal position error of 25.51%.It can be seen that UKF filter is better than EKF filter in smart car sensor fusion,which proves the validity and reliability of UKF filter algorithm in GNSS/INS sensor fusion.