In order to suppress the relative intensity noise (RIN) of broadband light source-based resonant fiber optic gyroscope (RFOG) and improve the accuracy of RFOG. The structure of the fiber optic resonant cavity sensitive element is analyzed and optimized. First, a simulation model is established based on the multi-loop interference superposition theory to explore the transmission characteristics of the broadband RFOG under different sensitive element structures. Next, based on the RFOG optical field theoretical model, the impact of coupler position on the accuracy of the gyroscope and the magnitude of the RIN is analyzed. Experimental results show that the asymmetric resonator with the fiber loop at the coupler's cross port achieves superior accuracy. And this structure suppresses relative intensity noise. By optimizing the sensitive element structure and using a fiber resonator with a diameter of 10 cm and a length of 220 meters, the angle random walk reached and the bias instability reached 0.002 (°)/h. In addition, a single-axis gyroscope prototype is built using a fiber loop with a diameter of 5 cm and a length of 300 meters. The prototype has a volume of 170 cm³. the angle random walk reachedand the bias instability reached 0.02(°)/h.

To improve the accuracy of Ultra Wide Band (UWB) localization in Non-Line-of-Sight (NLOS) scenarios, a NLOS recognition and localization algorithm based on credibility is proposed. This algorithm utilizes UWB real-time Channel Impulse Response (CIR) features and ranging values to identify Line-of-Sight (LOS) or NLOS through one-dimensional convolution neural network, and outputs the probability of LOS or NLOS. Then this probability is used to construct credibility. Based on credibility, base station selection and improved positioning algorithms are carried out. Weighted Least Squares and Taylor (WLS-Taylor) fusion algorithm based on credibility is designed. Static and dynamic measured data in various scenarios is collected to verify the performance of the proposed algorithm. The experimental results show that the proposed algorithm can effectively suppress the influence of NLOS on positioning results, and the average positioning error is less than 10 cm in NLOS environment. In environments with relatively severe NLOS, the positioning error of the proposed algorithm is reduced by 76.94 cm compared to the WLS algorithm based on distance weighting.

The scene matching and positioning of Unmanned aerial vehicles (UAVs) are prone to mismatching or even retrieval failure due to the differences in domain, observation angle and other factors between UAV images and satellite reference images. To address this issue, a rapid cross-source image retrieval method based on salient location features is proposed. Firstly, to solve the matching failure caused by scene and time differences between UAV images and reference images, a salient position feature extraction module is designed, which can extract more effective context information while reducing the computational complexity. Secondly, a label smoothing loss function is introduced to enhance the generalization ability of the model. Finally, a block-wise fine-tuning strategy is proposed to alleviate the overfitting problem of large models like vision transformer (ViT) under limited training data conditions. The experimental results show that the proposed method achieves 86.01% and 96.52% respectively in R@1 and R@5 on the DenseUAV dataset, and 76.04% in mAP, which is improved by 5.83%, 3.53% and 9.49% respectively compared with ViT-S. The retrieval time for a single image is 9.55 ms on the DenseUAV dataset, indicating the effectiveness of the proposed method in UAV cross-source scene matching.

To address the accuracy degradation of the Kalman filter (KF) algorithm defined in quaternion space under non-Gaussian noise, the advantages of mixture correntropy is utilized to handle such problem. A recursive quaternion mixture correntropy cost function is defined and the posterior estimation through fixed-point iteration is obtained, resulting in the maximum mixture correntropy quaternion KF (MMCQKF) algorithm. Additionally, the variational Bayesian method is introduced to adaptively update the nominal measurement noise variance matrix, leading to the adaptive MMCQKF, which further improves state estimation accuracy in complex scenarios. Simulation results for target tracking in challenging noise environments show that the root mean square error of position estimation using the proposed algorithm is reduced by approximately 53.2% compared to the maximum correntropy quaternion KF. Furthermore, integrated navigation experiments conducted in complex non-Gaussian noise environments reveal that the error in attitude angle, velocity, and position achieved by the proposed algorithm are reduced by 70.6%, 59.1% and 73.1%, respectively, compared to the maximum correntropy quaternion KF. Experiments demonstrate the significant improvement in estimation accuracy and adaptive capability of the proposed algorithm.

Traditional neural network algorithms are prone to consuming a long time and getting stuck in local optima when artificial intelligence methods are applied to geomagnetic navigation and positioning. To address these issues, a method for geomagnetic positioning of aircraft based on improved gradient-based genetic algorithm optimized extreme learning machine neural network (GGA-ELM) is proposed. The training efficiency is greatly improved based on the optimized ELM network and the risk of falling into local optimum is effectively reduced as well by introducing an elite reverse learning strategy into the traditional genetic algorithm. Some aeromagnetic data measured by drone are used for investigation. The experimental results show that the training time of the GGA-ELM model is significantly reduced compared with the CNN, BiLSTM and LSTM models. In addition, the localization error of the GGA-ELM model is about 4 m, and the localization time is 0.003 s. Compared with the ELM, GA-ELM, CNN, BiLSTM, RBF and LSTM models, based on the GGA-ELM method, the localization accuracy is improved by 86.6%, 115.9%, 417.8%, 187.6%, 216.5%, and 107.5%, respectively. The localization time is reduced up to 0.947 s. From the results, it is clearly seen that the proposed method has better positioning stability and higher accuracy on aircraft localization.

Regarding the shock failure of a micro-electro-mechanical system (MEMS) vibrating ring gyroscope (VRG) under high overload, structural dynamic response modeling and failure mechanism analysis are conducted. Based on vibration and elastic wave theories, a dynamic response model of gyroscopic structures to high-g shocks is established. Based on the established dynamic impact response model of the MEMS ring gyroscope, the adhesion and fracture failure mechanisms of the MEMS ring gyroscope are analyzed. The equilibrium displacement for adhesion failure and the sensitive location for fracture failure are derived, and the impact expression at the point of failure is obtained. Through high overload experiments, the impact amplitude and pulse width at the critical failure of the MEMS ring gyroscope a re determined. Raman spectroscopy is used to test the surface stress of the MEMS ring gyroscope after high overload application, and the stress-sensitive locations are found to be consistent with theoretical derivations.

In order to solve the problem of the degradation of the autonomous positioning accuracy of the land-based inertial/odometry dead reckoning (DR) system for long-term and large-range driving under satellite denial, an inertial/map matching error identification method based on global geometric feature sliding optimization is proposed. Firstly, a map matching algorithm based on the geometric features of the trajectory is designed to achieve the accurate matching of the DR trajectory and the road data of the electronic map. Secondly, after obtaining the accurate map matching results, based on the principle of similarity between the DR trajectory and the real trajectory, a DR error identification and compensation method based on the sliding optimization of global geometric features is proposed, which identified and compensated the position error and odometer scale coefficient error of the navigation system. The vehicle experiment results show that the proposed method can achieve high-precision autonomous positioning of vehicles under large-scale driving conditions, and the maximum horizontal positioning error throughout multiple large-scale long-distance (with driving distances all exceeding 160 km) on-board experiments is 12.76 m, compared with the traditional translational vector compensation and adjacent feature point identification and compensation methods, the maximum positioning error is reduced by 53.1% and 31.0% on average, the root mean square error is reduced by 50.0% and 39.0% on average, which verifies the effectiveness of the proposed method.

To address the issue of poor adaptability of traditional Kalman filters to nonlinear non-Gaussian measurement signals in relative navigation of non-cooperative spacecraft, which can lead to performance degradation or even divergence, a nonlinear filtering method based on α-divergence minimization (αKF) is proposed. Operating within the Bayesian estimation framework, this method achieves high-precision dynamic solution of relative position and velocity between the observing and observed spacecraft by optimizing posterior probability distribution estimation through α-divergence minimization. Simulation experiments demonstrate the robustness of the proposed method under both Gaussian and second-order Gaussian mixture models (GMM). Results indicate that under second-order GMM non-Gaussian noise conditions, the αKF-based algorithm achieves relative position estimation accuracy of 1.813 m and relative velocity precision of 0.022 m/s. Furthermore, parameter sensitivity analysis reveals the optimal range for divergence coefficient α to be 0.05~0.1, providing valuable reference for filter parameter configuration in complex noise scenarios.

Aiming at the problem of robot pose estimation bias and imperfect map construction caused by moving objects in dynamic scenes, an ORB-SLAM3 algorithm for dynamic scene optimization is proposed. Firstly, the dynamic object is detected by the improved YOLOv5s algorithm and the associated feature points are preliminarily removed. Then, the missing dynamic feature points are further filtered by combining LK optical flow tracking and epipolar geometric constraint analysis based on fundamental matrix, so as to improve the accuracy of environment perception and pose estimation. At the same time, the corresponding point cloud information is generated by filtering the key frames of dynamic information to realize the construction of 3D dense static map. The test results in indoor dynamic scenes show that compared with the traditional ORB-SLAM3, the absolute trajectory error and relative pose error of the proposed algorithm are reduced by 55.2% and 93.7% respectively in the office environment, and by 24.3% and 40.2% in the corridor scene, which verifies the robustness advantage of the proposed algorithm in dynamic scenes.

To address the challenge of accurately determining the environmental region of unmanned vehicles during seamless indoor-outdoor positioning, a regional recognition method for seamless indoor-outdoor localization is proposed. Firstly, a joint prediction model integrating particle swarm optimization-support vector machine (PSO-SVM) and hidden Markov model (HMM) is designed. Environmental feature data acquired by sensors serve as model inputs to generate regional recognition results. Secondly, three environmental models are introduced to describe the vehicle's operational environment, with corresponding measurement information selected based on the regional recognition outcomes. Finally, the regional transition probabilities are utilized to update the switching probabilities of the three environmental sub-models in the interactive multiple model (IMM) algorithm, thereby enhancing the accuracy of environmental region recognition and positioning precision for seamless indoor-outdoor navigation. The results of real-vehicle experiment show that the proposed joint recognition method achieves an accuracy of 98.09% in region recognition, representing improvements of 2.13% and 9.53% compared to using PSO-SVM or HMM alone. Further experiments indicate that the proposed seamless positioning method enhances the average positioning accuracy by 43.75% and 22.30% compared to the traditional federated Kalman filter (FKF) algorithm and IMM algorithm, respectively.