In this study, both adaptive and robust one is proposed, considering that the properties of a classic unscented Kalman filter (UKF) can be degraded severely by the outliers measured in the contamination distribution and the effects of time-varying noise. An adaptive and robust UKF approach on Gaussian process regression- assisted Variational Bayesian is proposed. The Variable Bayesian (VB) method is used to statistically approximate the time-varying noise due to the strong nonlinearity and inaccuracy of the system model. At the same time, Gaussian Process Regression (GPR) has the advantage of machine learning. The observation information trained by the GPR model performs real-time prediction of the sliding window. Based on the above two points, the output source of the measurement information is estimated to realize the robustness of the filter measurement update. When the noise is uncertain and there are outliers in the measurement information, simulation and SINS/GPS integrated navigation of actual tests were performed, respectively. The test outcomes indicate that the algorithm has better nonlinear estimation performance than the traditional method, the effectiveness of the filtering algorithm proposed is verified.

The accuracy and robustness of SINS/GNSS integrated navigation are crucial in autonomous driving navigation applications. However, under the conditions of urban high-rise buildings and roadside trees, as well as the limitations of the Global Navigation Satellite System (GNSS), the traditional robust algorithm design at the filtering level is difficult to meet the needs of the actual vehicle movement environment. We developed a GNSS-independent, inertial-based navigation technology called vehicle model-assisted inertial-based integrated navigation robust filtering algorithm. Without adding external navigation sensors, the vehicle motion characteristics are modeled, and dead reckoning observations are used to assist SINS status updates. The more flexible federal unscented Kalman filter (UKF) framework is designed considering vehicle kinematics and dynamics for nonlinear navigation systems. Our strategy will provide reliable observable physical quantities. The simulation test results show that the accuracy and robustness of the low-cost vehicle inertial-based integrated navigation system are improved. This new method reduces the heading angle error by 84% and the corresponding position error by 96% during GNSS outages, as demonstrated in this paper. The effectiveness of the algorithm proposed in the paper was verified.

Autonomous driving vehicle is a kind of unmanned vehicle on the ground. Navigation and positioning technology is becoming an indispensable part of practical applications, especially in intelligent transportation systems and military fields. In this study, multisource information fusion technology plays a necessary role in enhancing the quality and reliability of navigation and positioning. In the actual complex environment, the accuracy requirements of vehicle navigation and positioning in the case of nonlinear systems and non-Gaussian distribution are fully considered. In this article, a new adaptive sharing factor integrated navigation information fusion technology scheme is designed. From a geometric point of view, the VDM is studied, and new VDM observation information that fully considers the turning process of the vehicle is established. It combines strapdown inertial navigation system (SINS), global navigation satellite system (GNSS), and VDM multisource sensors, and a VDM and GNSS measurement model based on nonintegrity constraints is developed. The performance of the local filter is adaptively reflected by the F-norm adaptive information sharing factor algorithm established based on the variance. At the same time, the development process of a federated filter fault diagnosis algorithm based on unscented Kalman filter (UKF) is reported, which has fault diagnosis and information recovery functions. The independent characteristics of the subfilters are considered, and a fault-tolerant mechanism for state chi-square detection is established. Both GNSS and VDM are used as observation sources to improve the accuracy of low-cost SINS. This strategy is implemented on the basis of Carsim professional vehicle dynamics software. The evaluation test indicated that the algorithm has high navigation accuracy, high fault tolerance, and guaranteed system reliability, which is verified in this article.

In order to solve the problem that the spherical harmonic model has a large amount of calculation, especially in the background of aviation navigation, it is difficult to realize real-time gravity compensation, a gravity compensation method based on back-propagation neural network (BPNN) is proposed. Firstly, the highest order/degree spherical harmonic model EIGEN-6C4 is used to calculate the gravity vector information in the planning navigation area, which is regarded as the true value. The gravity disturbance information and the corresponding position information are used as the training data set of BPNN. Then, the gravity model based on BPNN is obtained by training, and the model is applied to the gravity compensation of INS. Finally, the effectiveness of the method is verified by experiments. The numerical simulation results show that the standard deviations of east-west, north-south gravity disturbances errors and gravity anomaly errors are 0.29 mGal, 0.35 mGal and 0.81 mGal, respectively. The vehicle experiment results show that the horizontal radial position performance of INS after gravity compensation is improved by more than 26.68%, which is basically consistent with the gravity compensation using the highest order/degree spherical harmonic model, and ensures the real-time performance of the algorithm.

Visual localization plays an essential role in the autonomous flight of Unmanned Aerial Vehicles (UAVs) especially for the Global Navigation Satellite System (GNSS) denied environments. Existing aerial-based visual localization methods mainly focus on eliminating image variance between database map and captured frames. However, these is a lack of public dataset and baseline for method comparisons, which impedes the development of aerial-based visual localization. To address this issue, we construct AerialVL, a large-scale dataset, which is collected using UAV flying at different altitudes, along various routes, and during diverse time periods. AerialVL consists of 11 image sequences covering approximately 70 km of trajectory and includes a reference satellite image database corresponding to the flight area. Leveraging AerialVL, we perform thorough evaluations on various mainstream solutions designed for aerial-based visual localization for the first time. This evaluation encompasses visual place recognition, visual alignment localization and visual odometry, serving as comparison baselines. Furthermore, we present a general aerial-based visual localization framework, which unifies various methods and integrates them into a modular architecture. We note that across all flight trajectories, the proposed framework achieves higher localization accuracy and robustness against the existing methods.

The gravity gradient is the second derivative of gravity potential. A gravity gradiometer can measure the small change of gravity at two points, which contains more abundant navigation and positioning information than gravity. In order to solve the problem of passive autonomous, long-voyage, and high-precision navigation and positioning of submarines, an aided navigation method based on strapdown gravity gradiometer is proposed. The unscented Kalman filter framework is used to realize the fusion of inertial navigation and gravity gradient information. The performance of aided navigation is analyzed and evaluated from six aspects: long voyage, measurement update period, measurement noise, database noise, initial error, and inertial navigation system device level. When the parameters are set according to the benchmark parameters and after about 10 h of simulation, the results show that the attitude error, velocity error, and position error of the gravity gradiometer aided navigation system are less than 1 arcmin, 0.1 m/s, and 33 m, respectively.