Precise point positioning (PPP) has received much attention in recent years for its low cost, high accuracy, and global coverage. Nowadays, PPP with ambiguity resolution and atmospheric augmentation is widely regarded as PPP-RTK (real-time kinematic), which weakens the influence of the long convergence time in PPP and regional service coverage in RTK. However, PPP-RTK cannot work well in urban areas due to limitations of non-line-of-sight (NLOS) conditions. Inertial navigation systems (INS) and vision can realize continuous navigation but suffer from error accumulation. Accordingly, the integration model of multi-GNSS (global navigation satellite system) and PPP-RTK/INS/vision with a cascading Kalman filter and dynamic object removal model was proposed to improve the performance of vehicle navigation in urban areas. Two vehicular tests denoted T01 and T02 were conducted in urban areas to evaluate the navigation performance of the proposed model. T01 was conducted in a relatively open-sky environment and T02 was collected in a GNSS-challenged environment with many obstacles blocking the GNSS signals. The positioning results show that the dynamic object removal model can work well in T02. The results indicate that multi-GNSS PPP-RTK/INS/vision with a cascading Kalman filter can achieve a positioning accuracy of 0.08 m and 0.09 m for T01 in the horizontal and vertical directions and 0.83 m and 0.91 m for T02 in the horizontal and vertical directions, respectively. The accuracy of the velocity and attitude estimations is greatly improved by the introduction of vision.

Intelligent vehicle (IV) navigation is a key component of intelligent transportation systems (ITS). To enhance their performance in positioning, navigation and timing (PNT), usually inputs from the exteroceptive sensors, e.g., camera (images) or light detection and ranging (LiDAR) (point clouds), can be used to identify the environment and objects, or help other sensors with signal quality assessment. However, the measurements from the interoceptive sensors, e.g., inertial measurement unit (IMU) and wheel odometer (ODO), are rarely considered, while they are more resilient to the environment changes and commonly equipped within the modern vehicles. Integrating the global navigation satellite system (GNSS) precise point positioning (PPP) algorithm and centered on the inertial navigation system (INS), this paper proposes a tightly coupled (TC) GNSS PPP/INS/ODO integration algorithm that utilizes long short-term memory (LSTM) to predict non-holonomic constraint (NHC) components. To further assess the online prediction quality without ground truth, a novel cross-validation method is proposed, also to adjust the corresponding measurement variance. Results of field tests show that, compared with traditional GNSS PPP/INS/ODO system, the proposed algorithm can achieve an average root mean square (RMS) improvement of (16.05%, 16.03%, 16.06%) and a maximum (MAX) improvement of (22.10%, 27.57%, 29.90%) in positioning along the north, east and down (NED) directions. More importantly, the proposed method can effectively mitigate the degradation in positioning performance caused by erroneous constraints under complex motion conditions.

Motion averaging (MA) offers an efficient and mostly linear means for estimating image sets pose and provides a reliable initialization for large-scale structure-from-motion (SfM) pipelines. Nonetheless, MA, which comprises rotation and translation averaging (RA & TA, respectively), can be severely affected by the presence of outliers that can greatly degrade the performance of the entire optimization process or even lead to the divergence of the SfM solution. While robust loss functions have been applied to mitigate outliers effect, their performance depends heavily on the choice of parameters and requires manual tuning and prior knowledge of the residual distribution. To make MA more robust, we enhance it by incorporating an adaptive robust kernel that automatically adjusts its parameter to the residual distribution. This adaptive behavior balances robustness and sensitivity, removing the need for manual parameter tuning. In addition, to address the ill-posed nature of TA, we adopt a global positioning framework that jointly estimates the camera and 3D point positions. Experimental results show how the adaptive robust kernel consistently outperforms state-of-the-art fixed-parameter functions. It improves accuracy in both RA and global positioning, particularly in scenes with high levels of noise or outliers. These results demonstrate the effectiveness of adaptive robust kernels for improving the reliability and generalization of MA pipelines in challenging reconstruction scenarios.

Visual-inertial odometry (VIO) plays a key role in modern navigation and mapping systems. For their successful integration, an initialization phase, in which IMU-related bias factors are estimated, becomes a fundamental step. Without one, the subsequent nonlinear estimation of the platform pose may fail to converge or completely diverge. As reliance on visual and inertial information may exhibit instability due to error accumulation with time, incorporating absolute positioning information through global navigation satellite system (GNSS) measurements, may enhance its robustness and accuracy. Accordingly, GNSS and visual-inertial initialization frameworks have been receiving growing attention in recent years where current strategies tend to follow a loosely-coupled formulation that first initializes the VIO trajectory, and then aligns it with GNSS measurements. Such strategies are multi-stage, nonlinear, and computationally expensive, motivating us to introduce an alternative framework in which GNSS position is integrated with the raw visual-inertial measurements to form absolute translation constraints. Accordingly, we achieve a closed-form, linear and globally consistent drift-free solution which is computationally efficient and requires neither 3D reconstruction nor nonlinear refinement, as common approaches do. Testing our initialization formulation on benchmark multi-sensor datasets, results show that we outperform current baselines while exhibiting robustness in challenging scenarios.