During the execution of autonomous tasks within sheltered space environments, unmanned vehicles demand highly precise and seamless continuous positioning capabilities. While the existing visual–inertial-based positioning methods can provide accurate poses over short distances, they are prone to error accumulation. Conversely, radio-based positioning techniques could offer absolute position information, yet they encountered difficulties in sheltered space scenarios. Usually, three or more base stations were required for localization. To address these issues, a binocular vision/inertia/ultra-wideband (UWB) combined positioning method based on factor graph optimization was proposed. This approach incorporated UWB ranging and positioning information into the visual–inertia system. Based on a sliding window, the joint nonlinear optimization of multi-source data, including IMU measurements, visual features, as well as UWB ranging and positioning information, was accomplished. Relying on visual inertial odometry, this methodology enabled autonomous positioning without the prerequisite for prior scene knowledge. When UWB base stations were available in the environment, their distance measurements or positioning information could be employed to institute global pose constraints in combination with visual–inertial odometry data. Through the joint optimization of UWB distance or positioning measurements and visual–inertial odometry data, the proposed method precisely ascertained the vehicle’s position and effectively mitigated accumulated errors. The experimental results indicated that the positioning error of the proposed method was reduced by 51.4% compared to the traditional method, thereby fulfilling the requirements for the precise autonomous navigation of unmanned vehicles in sheltered space.

None-Line-of-Sight (NLOS) propagation of Ultra-Wideband (UWB) signals leads to a decrease in the reliability of positioning accuracy. Therefore, it is essential to identify the channel environment prior to localization to preserve the high-accuracy Line-of-Sight (LOS) ranging results and correct or reject the NLOS ranging results with positive bias. Aiming at the problem of the low accuracy and poor generalization ability of NLOS/LOS identification methods based on Channel Impulse Response (CIR) at present, the multilayer Convolutional Neural Networks (CNN) combined with Channel Attention Module (CAM) for NLOS/LOS identification method is proposed. Firstly, the CAM is embedded in the multilayer CNN to extract the time-domain data features of the original CIR. Then, the global average pooling layer is used to replace the fully connected layer for feature integration and classification output. In addition, the public dataset from the European Horizon 2020 Programme project eWINE is used to perform comparative experiments with different structural models and different identification methods. The results show that the proposed CNN-CAM model has a LOS recall of 92.29%, NLOS recall of 87.71%, accuracy of 90.00%, and F1-score of 90.22%. Compared with the current relatively advanced technology, it has better performance advantages.

Visual-based positioning technology plays a pivotal role in spatial artificial intelligence, facilitating precise perception and comprehension of the physical world for robotic platforms and augmented reality devices. In this study, we propose a binocular camera-based method for spatial localization of targets using CNN for instance segmentation while simultaneously providing target location information. The method encompasses image acquisition and correction, target recognition and segmentation, and stereo matching, among other components. Building upon this foundation, we introduce a pedestrian recognition segmentation network model with an attention mechanism. To accurately locate the target, we employ a multi-feature fusion feature point extraction and matching algorithm that combines edge information with semantic information. Finally, our proposed method is evaluated for dynamic pedestrian targets in indoor environments, achieving a horizontal positioning error of less than 0.25 m.

LiDAR SLAM relies on reliable geometric constraints to estimate sensor motion and maintain consistent maps in complex three-dimensional environments. Planar features are commonly used for LiDAR registration, but repeatedly fitting local planes from neighboring points brings extra computation and may be sensitive to noisy observations. Voxel-plane maps address this issue by storing planar structures in voxel cells, yet most existing methods still construct planes and assign points to voxels according to the nominal point coordinates. When LiDAR measurement noise and pose prediction errors are ignored, plane parameters may be biased and points may be associated with unsuitable voxels. This paper presents UVP-LIO, an uncertainty-aware voxel-plane mapping method for LiDAR-inertial odometry. The measurement uncertainty of each LiDAR point and the uncertainty from state estimation are jointly propagated to the world frame to obtain a point-wise covariance. This covariance is then used in uncertainty-aware voxel association and covariance-weighted incremental plane updating. Plane thickness is further introduced to weight point-to-plane residuals during registration. Experiments in a LiDAR-only configuration on KITTI and in a LiDAR-inertial configuration on M3DGR show that UVP-LIO improves trajectory consistency and mapping robustness, especially in scenes with weak or ambiguous geometric constraints, while maintaining real-time performance.