Research On Fusion And Evoluation Method Of Visiual Inertial Navigation

With the rapid development of computer science and sensor technology,there has an explosive growth of concern in high-precision positioning services,such as driverless car,Internet of Things,UAV positioning,augmented reality,etc.The visual inertial fusion navigation system stands out among many positioning technologies with its excellent performance,low cost and wide application scenarios,and has become a research hotspot in recent years.In this paper,some key methods of visual inertial fusion navigation are studied,and related systems are designed and implemented.This paper mainly studies two kinds of visual inertial fusion navigation methods.One is a visual inertial fusion navigation method based on loosely coupled filter.The other is the visual inertial fusion navigation method based on tightly coupled nonlinear optimization.The main research work of this paper is summarized as follows:Firstly,this paper studies a visual inertial fusion navigation method based on loosely coupled filtering.The method adopts the Extended Kalman filter to fuse visual odometry and inertial information,has low computational complexity and high operating efficiency,and is very suitable for indoor positioning of mobile smart devices.This paper implements a visual inertial fusion navigation system based on this method,and designs experiments to verify.Secondly,this paper studies a visual inertial fusion navigation method based on tightly coupled nonlinear optimization,and innovatively proposes a robust system initialization method.In this paper,we designs and implements the visual inertial fusion navigation system based on this method,and designs some experiments to verify.The result proves that this method can achieve a fairly high positioning accuracy and is suitable for scenes with high precision requirements.Finally,this paper establishes a reliable visual inertial Dataset,and innovatively proposes an internal parameter calibration method for inertial measurement unit(IMU)using the high-precision motion capture system.