Vehicle Fusion Positioning Based On IMU And GNSS

In recent years,the research of driverless vehicles has become a field of high attention.Advanced driver assistance systems are also widely used in modern vehicles.A large number of different sensors are also equipped with these systems,such as global navigation satellite system(GNSS)receiver,inertial measurement Unit(IMU),odometer,etc.While these sensors improve vehicle positioning,there are also many problems.Although the vehicle positioning technology based on auxiliary systems such as Lidar and visual odometer has made great progress,its positioning accuracy and reliability are still insufficient,and the cost is high,which can not be applied to the general intelligent vehicle.Moreover,in complex urban environments,GNSS signals cannot be effectively positioned due to environmental interference,which leads to unreliable positioning accuracy and fails to meet vehicle positioning requirements.To solve the above problems,this thesis focuses on IMU/GNSS fusion positioning technology,and mainly conducts the following researches:In this thesis,the cost of auxiliary system is studied,and the effectiveness of an ultra-low cost micro-electromechanical(MEMS)MPU-6000 IMU,which is widely used in various fields but not in the field of automotive navigation,is verified in the application of vehicle positioning technology.In this experiment,we built three systems:MPU-6000/U-blox(low-cost IMUs and low-cost GNSS),MPU-6000/ Ekinox(lowcost IMUs and mid-range GNSS),and Ekinox/ Ekinox(mid-range IMUs and midrange GNSS).The first two systems were designed to test the performance of the MPU-6000 IMUs in combination with higher quality GNSS sensors.The performance analysis of the MPU-6000 IMU is mainly divided into two stages: static analysis and kinematic analysis.Static analysis mainly uses Allen variance to analyze the error and test the normality.The kinematic analysis mainly analyzes and compares the trajectory,attitude,position,speed and error of the latter two systems,and compares and analyzes the performance of the latter two systems.The experimental results preliminarily show that the proposed low-cost IMU based on-board fusion positioning is feasible,that is,while reducing the cost of sensors,it can obtain the positioning effect with high precision.Aiming at the problem that GNSS signals cannot be effectively positioned due to environmental interference in the sheltered environment,and the inertial navigation error cannot be corrected for a long time without external assistance,which leads to the serious divergence of positioning accuracy,In this thesis,we propose a method to reduce the errors of inertial navigation and odometer by using incompleteness constraints and state-aware pseudo-measurements(e.g.,zero velocity and zero angular velocity)that are updated during periodic stops.In this thesis,a combination of many update type solutions was performed for each test run,and then the error estimation with or without zero type update was tested,and compared with the positioning method based on GPS.It was proved that the positioning algorithm based on periodic zero type constraint proposed in this thesis can significantly reduce the growth rate of vehicle navigation errors,and can obtain a high precision positioning effect.At the same time,it is proved that vehicle slip can be detected by using threshold and slip ratio estimation on speed residual.This thesis has 39 figures,12 tables and 80 references.