Research On Vehicle Inertial/Vision/GNSS Integrated Seamless Navigation Algorithm

At present,the development of unmanned vehicle and intelligent traffic puts forward higher requirements on the accuracy,reliability and seamless of vehicle navigation.The single navigation method always has its inherent defects.In order to achieve seamless navigation in multiple scenarios change,integrated navigation with multiple sensors is generally needed.The main research work and innovation points of this paper are as follows:(1)An improved MSCKF algorithm based on state transformation Kalman filter is proposed.The state equation and observation equation of the traditional MSCKF algorithm are derived in detail.In view of the inconsistency in the MSCKF algorithm,ST-EKF is used to replace the traditional EKF,which can avoid introducing the wrong observability for the unobservable state and improve the consistency and navigation accuracy of the navigation algorithm.(2)A structured environment constraint and motion constraint algorithm are proposed.Aiming at the problem of sparse and uneven distribution of feature points in inertial/visual integrated navigation,the influence of feature point density and distribution on navigation results was studied.In order to eliminate the adverse effects,the algorithm of structured environment constraint and motion constraint is designed to realize the seamless navigation when the visual information is partially or completely invalid.(3)An inertial/vision/GNSS seamless navigation algorithm is designed.The navigation system can maintain high accuracy when the satellite information or visual information is limited,and can seamlessly connect the algorithm operation and navigation output when the scene is switched.