Research On Key Technologies Of GNSS/INS/VO Integrated Navigation System

With the rapid development of science and technology,a single navigation system can no longer meet people’s requirements for higher navigation accuracy,robustness and reliability of the carrier.Therefore,the integrated navigation technology should be used to combine the navigation systems with different characteristics organically,which can realize the complementary advantages of each navigation system.Then it can improve the overall performance of the navigation system.The thesis investigates the key technologies of the GNSS/INS/VO integrated navigation system under the background of vehicle navigation.The main study work and innovations are as follows:(1)Aiming at the problem that the existing adaptive square-root cubature Kalman filter(ASCKF)algorithm based on the moving window method has low filtering accuracy,a weighted ASCKF(WASCKF)algorithm is proposed in this thesis.When using the moving window method to estimate the maximum likelihood of innovation covariance matrix,the proposed algorithm dynamically sets the corresponding weights are set dynamically according to the usefulness of the innovation vectors at different times in the window,which can enhance the utilization of the useful innovation vectors.Thereby,it can obtain an accurate estimation of the measurement noise statistics and improve the filtering accuracy of GNSS/INS integrated navigation system.Secondly,a modified ASCKF(MASCKF)algorithm is proposed to solve the problem that the window width of the existing ASCKF algorithm is difficult to determine.The algorithm adopts an adaptive filtering framework with a plurality of moving window estimators of different widths.Then the weight coefficient is used to optimize the innovation covariance matrix,which can reduce the filtering error when the window width is not properly selected.The experimental and simulation results show that the navigation performance of the proposed algorithms are better than SCKF and the existing ASCKF algorithm when there is uncertainty in the measurement noise statistics.(2)Aiming at the problems of the existing strong tracking SCKF(STSCKF)algorithm with low filtering accuracy and high calculation complexity,this thesis proposes an improved STSCKF algorithm.In this algorithm,a simple measurement updating process based on singular value decomposition is used to improve the computational efficiency and numerical stability of the algorithm.Then,according to the orthogonality principle of the prediction residual sequence,a method for calculating the suboptimal fading factor is derived,which is simple and does not need the Jacobian matrix.Meanwhile,the method of the hypothesis test is used to detect the uncertainty of the process model.Only when the uncertainty of the process model is detected,the suboptimal fading factor is introduced into the square root matrix of the covariance matrix of state prediction,which effectively solves the problem of the loss of filtering accuracy in the existing STSCKF algorithm.Finally,the effectiveness of the proposed algorithm is verified by numerical simulation and GNSS/INS integrated navigation experiments.(3)Aiming at the problem that large computation and poor navigation performance in complex environment in the VIO integrated navigation system,a robust adaptive VIO integrated navigation algorithm based on fuzzy logic is proposed.In this algorithm,the epipolar geometry and trifocal tensor geometry relationship between different images are used as the measurement models of the VIO navigation system,which can avoid calculating the three-dimensional position of feature points and avoid the problem of large complexity such as multi-state constraint Kalman filter.In addition,aiming the problem that the navigation accuracy of the existing VIO integrated navigation system based on cubature multi-state constraint Kalman filter(CMSCKF)algorithm decreases under the abnormal measurement,the T-S fuzzy logic is used to obtain the scalar factor which can dynamically adjust the filter gain and improve the robustness of the VIO integrated navigation system.Finally,the publicly available KITTI data set is used to verify the proposed algorithm.The experimental results show that compared with the existing CMSCKF algorithm,the proposed algorithm improves the positioning accuracy and attitude accuracy by 44.93% and 20.32% respectively,which effectively improves the navigation performance and adaptive ability of the VIO integrated navigation system.(4)A GNSS/INS/VO integrated navigation system with adaptive federated filtering architecture is designed.In order to improve the navigation performance of the navigation system,when the master filter allocates information for the local filters,a method for adaptively adjusting the information allocation factor according to the navigation performance of the local filters is proposed.In addition,in order to solve the problem that the abnormal measurement of the local filters in the reset GNSS/INS/VO integrated navigation will pollute the performance of the whole navigation system,a detection algorithm is designed to detect the abnormal measurement of the local filter,which can avoid the impact of the abnormality on the navigation system.Finally,the experiments based on the vehicle-based KITTI dataset show that the proposed algorithm can not only provide reliable navigation results in the GNSS-denied environment but also improve the navigation performance and robustness of the GNSS/INS/VO integrated navigation system.