In urban areas, the accuracy and reliability of global navigation satellite systems (GNSS) vehicle positioning decline due to substantial non-line-of-sight (NLOS) signals and multipath effects. Recently, positioning enhancement approaches with supervised GNSS signal type classification based on 3D building model-aided labelling have received widespread attention. Despite the reduced computing costs and improved real-time performance, the strict requirements of 3D building models on accuracy and timeliness limit the popularization of the technology to some extent. Meanwhile, the diversity of anomalous observation sources is beyond the reach of NLOS/multipath detection methods. This paper attempts to construct an alternative framework for quality identification of GNSS observations combining clustering-based anomaly detection and supervised classification, in which the hierarchical density-based spatial clustering of applications with noise (HDBSCAN) algorithm is used to label the offline dataset as normal and anomalous observations without the aid of 3D building models, and the supervised classifier in the online system learns the classification rule for real-time anomaly detection. The experimental results based on the measured vehicle GPS/BeiDou data show that after excluding anomalous observations the single point positioning accuracy of the offline dataset is improved by 87.0%, 45.9%, and 69.6% in the east, north, and up directions, respectively, and the positioning accuracy of two online datasets is improved by 48.4%/45.7%, 39.6%/63.3%, and 49.6%/49.1% in the three directions. Through a large number of comparative experiments and discussion on key issues, it is certified that the proposed method is highly feasible and has great potential in the practical application of urban GNSS vehicle positioning.

An inter-system differencing model between two Global Navigation Satellite Systems (GNSS) enables only one reference satellite for all observations. If the associated differential inter-system biases (DISBs) are priori known, double-differenced (DD) ambiguities between overlapping frequencies from different GNSS constellations can also be fixed to integers. This can provide more redundancies for the observation model, and thus will be beneficial to ambiguity resolution (AR) and real-time kinematic (RTK) positioning. However, for Global Positioning System (GPS) and the regional BeiDou Navigation Satellite System (BDS-2), there are no overlapping frequencies. Tight combination of GPS and BDS needs to process not only the DISBs but also the single-difference ambiguity of the reference satellite, which is caused by the influence of different frequencies. In this paper, we propose a tightly combined dual-frequency GPS and BDS RTK positioning model for medium baselines with real-time estimation of DISBs. The stability of the pseudorange and phase DISBs is analyzed firstly using several baselines with the same or different receiver types. The dual-frequency ionosphere-free model with parameterization of GPS-BDS DISBs is proposed, where the single-difference ambiguity is estimated jointly with the phase DISB parameter from epoch to epoch. The performance of combined GPS and BDS RTK positioning for medium baselines is evaluated with simulated obstructed environments. Experimental results show that with the inter-system differencing model, the accuracy and reliability of RTK positioning can be effectively improved, especially for the obstructed environments with a small number of satellites available.

Traditional Simultaneous Localization and Mapping (SLAM) (with loop closure detection), or Visual Odometry (VO) (without loop closure detection), are based on the static environment assumption. When working in dynamic environments, they perform poorly whether using direct methods or indirect methods (feature points methods). In this paper, Dynamic-DSO which is a semantic monocular direct visual odometry based on DSO (Direct Sparse Odometry) is proposed. The proposed system is completely implemented with the direct method, which is different from the most current dynamic systems combining the indirect method with deep learning. Firstly, convolutional neural networks (CNNs) are applied to the original RGB image to generate the pixel-wise semantic information of dynamic objects. Then, based on the semantic information of the dynamic objects, dynamic candidate points are filtered out in keyframes candidate points extraction; only static candidate points are reserved in the tracking and optimization module, to achieve accurate camera pose estimation in dynamic environments. The photometric error calculated by the projection points in dynamic region of subsequent frames are removed from the whole photometric error in pyramid motion tracking model. Finally, the sliding window optimization which neglects the photometric error calculated in the dynamic region of each keyframe is applied to obtain the precise camera pose. Experiments on the public TUM dynamic dataset and the modified Euroc dataset show that the positioning accuracy and robustness of the proposed Dynamic-DSO is significantly higher than the state-of-the-art direct method in dynamic environments, and the semi-dense cloud map constructed by Dynamic-DSO is clearer and more detailed.

Due to the massive multipath effects and non-line-of-sight (NLOS) signal receptions, the accuracy and reliability of GNSS positioning solution can be severely degraded in a highly urbanized area, which has a negative impact on the performance of GNSS/INS integrated navigation. Therefore, this paper proposes a multipath/NLOS detection method based on the K-means clustering algorithm for vehicle GNSS/INS integrated positioning. It comprehensively considers different feature parameters derived from GNSS raw observations, such as the satellite-elevation angle, carrier-to-noise ratio, pseudorange residual, and pseudorange rate consistency to effectively classify GNSS signals. In view of the influence of different GNSS signals on positioning results, the K-means clustering algorithm is exploited to divide the observation data into two main categories: direct signals and indirect signals (including multipath and NLOS signals). Then, the multipath/NLOS signal is separated from the observation data. Finally, this paper uses the measured vehicle GNSS/INS observation data, including offline dataset and online dataset, to verify the accuracy of signal classification based on double-differenced pseudorange positioning. A series of experiments conducted in typical urban scenarios demonstrate that the proposed method could ameliorate the positioning accuracy significantly compared with the conventional GNSS/INS integrated navigation. After excluding GNSS outliers, the positioning accuracy of the offline dataset is improved by 16% and 85% in the horizontal and vertical directions, respectively, and the positioning accuracy of the online dataset is improved by 21% and 41% in the two directions. This method does not rely on external geographic information data and other sensors, which has better practicability and environmental adaptability.

The regional constellation of the BeiDou navigation satellite system (BDS) has been providing continuous positioning, navigation and timing services since 27 December 2012, covering China and the surrounding area. Real-time kinematic (RTK) positioning with combined BDS and GPS observations is feasible. Besides, all satellites of BDS can transmit triple-frequency signals. Using the advantages of multi-pseudorange and carrier observations from multi-systems and multi-frequencies is expected to be of much benefit for ambiguity resolution (AR). We propose an integrated AR strategy for medium baselines by using the combined GPS and BDS dual/triple-frequency observations. In the method, firstly the extra-wide-lane (EWL) ambiguities of triple-frequency system, i.e., BDS, are determined first. Then the dual-frequency WL ambiguities of BDS and GPS were resolved with the geometry-based model by using the BDS ambiguity-fixed EWL observations. After that, basic (i.e., L1/L2 or B1/B2) ambiguities of BDS and GPS are estimated together with the so-called ionosphere-constrained model, where the ambiguity-fixed WL observations are added to enhance the model strength. During both of the WL and basic AR, a partial ambiguity fixing (PAF) strategy is adopted to weaken the negative influence of new-rising or low-elevation satellites. Experiments were conducted and presented, in which the GPS/BDS dual/triple-frequency data were collected in Nanjing and Zhengzhou of China, with the baseline distance varying from about 28.6 to 51.9 km. The results indicate that, compared to the single triple-frequency BDS system, the combined system can significantly enhance the AR model strength, and thus improve AR performance for medium baselines with a 75.7% reduction of initialization time on average. Besides, more accurate and stable positioning results can also be derived by using the combined GPS/BDS system.

The distribution of total electron content (TEC) in the ionosphere is irregular and complex, and it is hard to model accurately. The polynomial (POLY) model is used extensively for regional ionosphere modeling in two-dimensional space. However, in the active period of the ionosphere, the POLY model is difficult to reflect the distribution and variation of TEC. Aiming at the limitation of the regional POLY model, this paper proposes a new ionosphere modeling method with combining the support vector machine (SVM) regression model and the POLY model. Firstly, the POLY model is established using observations of regional continuously operating reference stations (CORS). Then the SVM regression model is trained to compensate the model error of POLY, and the TEC SVM-P model is obtained by the combination of the POLY and the SVM. The fitting accuracies of the models are verified with the root mean square errors (RMSEs) and static single-frequency precise point positioning (PPP) experiments. The results show that the RMSE of the SVM-P is 0.980 TECU (TEC unit), which produces an improvement of 17.3% compared with the POLY model (1.185 TECU). Using SVM-P models, the positioning accuracies of single-frequency PPP are improved over 40% compared with those using POLY models. The SVM-P is also compared with the back-propagation neural network combined with POLY (BPNN-P), and its performance is also better than BPNN-P (1.070 TECU).