Research On VIO System Based On IMU Pre-integration And Nonlinear Optimization

Visual Inertial Odometry(VIO)realizes positioning and navigation of mobile robot by fusing visual and inertial measurement unit(IMU)data.With its advantages of strong complementarity,high positioning accuracy and good robustness,VIO has gradually become a common form of integrated navigation system,however,due to the large amount of visual information,multi-sensor data fusion and computational complexity,there are many research challenges in this navigation method.Therefore,this paper studies the visual-inertial integrated navigation system,and designs an embedded,real-time binocular VIO.Firstly,this paper introduces the theoretical basis of navigation,and expounds the positioning principle of camera and the state estimation method of VIO;Secondly,the binocular visual odometry is studied,with emphasis on the position and attitude estimation method of visual odometry based on feature method;finally,the VIO navigation system is designed,and the key technologies are studied,including IMU data pre-integration,Local map creation and maintenance,as well as back-end nonlinear optimization.Among them,IMU pre-integration and key frame technology are used to solve the problem of asynchronization between vision and IMU data.In the selection of key frames,IMU pre-integration is proposed as a priori to judge key frames by motion increment.In the part of non-linear optimization,aiming at the influence of IMU noise and deviation variation,the deviation of IMU is optimized as a state variable.Moreover,in order to reduce the dimension of road signs in local maps,only the inverse depth is optimized,and the depth information obtained by binocular vision is used as the observation of the depth of road signs,and more accurate depth is obtained than that obtained by mono-vision inertial combination.Information.In order to verify the design of the VIO,this paper designs an embedded binocular VIO system,which can achieve accurate real-time positioning and navigation.In the experimental analysis part,the navigation performance of the system is evaluated,and the validity and reliability of the VIO navigation system are verified.