This work aims to improve the low accuracy caused by interference information during real-time weld surface detection. Firstly, a weld surface dataset with 7580 pictures containing 8 types of defects was established. Secondly, an improved detection model named Yolo-MSAPF was designed based on Yolo-v5 model and verified by the self-established database. Finally, a real-time detection system was built to analyze the performance of the detection model in an industrial environment. The design principle of the Yolo-MSAPF is to eliminate interference information but enhance necessary features in each scale as much as possible by multi-scale alignment fusion (MSAF) with parallel features filtering (PFF) modules (i.e., MSAPF strategy). For the MSAF, not only the accuracy but the richness of fused features are guaranteed by aligning features at one level to fuse all other scales. After that, the fused features in each scale were individually filtered out in parallel spatial and channel in the PFF module. The results show that rate for the images with missing detection for 8 types of defects sharply drops from 21.47% to 1.68% when the MSAPF strategy is used. Moreover, it is worth noting that the mAP @0.5 of the Yolo-MSAPF model reaches 95.3%, which is similar to the Yolo-v7 model, while the number of parameters is reduced by 30.1%, compared with the baseline model. Additionally, the great ability to screen out unqualified weld is also verified in the industrial environment. Soon, code and dataset will be available at https://github.com/Luckycat518/Yolo-MSAPF.

The mitigation of the drift of the odometer and the ghost trail effect in the map is a challenge for simultaneous localization and mapping (SLAM). Facing the challenge, a novel method, called LIWOM-GD, is proposed via factor graph optimization, which tightly fuses LiDAR, IMU, and wheel encoder measurements. Firstly, the front-end module provides undistorted LiDAR scans, the pose states to be optimized, and the LiDAR-Inertial (LI) odometry factor. Secondly, the point cloud and pre-integration processing module implements ground segmentation, IMU, and encoder pre-integration to form the ground plane, IMU, and encoder factors. Furthermore, an enhanced dynamic point removal method is integrated into the processing module. Thirdly, a typical loop closure detection method is used to obtain a loop closure factor. Finally, all the above factors are added into the factor graph to obtain optimized results based on the GTSAM library. The proposed SLAM system tightly fuses three sensor measurements based on the factor graph, adding global plane constraint and dynamic points removal module based on spatiotemporal normals. Experimental tests are conducted on both benchmark and self-collected datasets. The results demonstrate that our method not only achieves more robust localization and mapping in degraded environments but also effectively implements dynamic point removal compared with the state-of-the-art SLAM methods.

In this letter, we propose a novel Non-Line-of-Sight (NLOS) identification and error-mitigation method for dynamic object positioning and ultra-wideband (UWB) ranging. By utilizing inverse estimation on known Anchor Points (APs) and improved robust unscented Kalman filter (IRUKF), while fusing Gyroscope and Accelerometer data, the proposed technology identifies and compensates for NLOS occlusions between tag and APs, reducing positioning errors. The approach has been verified through simulation and experiment, with identification precision of 97.02%. After mitigating errors, substantial error reductions of 91.80% and 98.90% were observed in LOS and NLOS situations, respectively. Moreover, the developed IRUKF effectively minimizes mislocalization by 50.48% in harsh scenarios.

Ultra-wideband (UWB) technology is extensively applied in indoor high-precision localization scenarios. However, barriers along the radio signal path can lead to Non-Line-of-Sight (NLOS) propagation, thereby reducing positioning reliability. Therefore, identifying NLOS conditions and mitigating the associated errors is essential. In this paper, a novel NLOS detection and mitigation approach is introduced, leveraging the parallel Spatio-Temporal feature fusion network (PSTFFN) and Spatio-Temporal attention regression network (STARN). We utilize continuous wavelet transform to convert the one-dimensional channel impulse response (CIR) signal into a two-dimensional time-frequency diagram of CIR image data. By incorporating an attention mechanism and handcrafted features, PSTFFN enhances its ability to differentiate between scenarios with similar features. Based on the NLOS classification results, the proposed STARN model, tailored to specific scenarios, further mitigates NLOS errors. Experimental results indicate that the proposed method performs highly in both binary and multi-class NLOS classification, with accuracy rates exceeding 96.74%. Moreover, using this approach, the root mean square error (RMSE) and mean absolute error (MAE) of distance measurements are reduced from 168.04 cm and 106.86 cm to 7.54 cm and 7.78 cm, respectively. In real-world indoor positioning experiments, this method improves positioning accuracy by over 81.86% and achieves real-time response performance.

The essence of the difficulties for weld surface detection is that there is a lot of interference information during detection. This study aims to enhance the detection accuracy while keeping great deployment capabilities of a detection model for weld surface defects. To achieve this goal, an improved Yolo-graph convolution head (GCH) model is proposed based on the stable and fast Yolo-v5. The improvements primarily involve introducing a graph convolution network combined with a self-attention mechanism in the head part (i.e. GCH). This component focuses on improving the insufficient recognition capability of convolutional neural networks for similar defects in complex environments. Furthermore, to address the presence of potentially ambiguous samples in complex welding environments, the label assignment strategy of simOTA is implemented to optimize the anchor frame. Additionally, a streamlined structure, aiming to improve model detection speed while minimizing performance impact, has been designed to enhance the applicability of the model. The results demonstrate that the cooperation of GCH and simOTA significantly improves the detection performance while maintaining the inference speed. These strategies lead to a 2.5% increase in mAP@0.5 and reduce the missing detection rates of weld and 8 types of defects by 32.9% and 84.1% respectively, surpassing other weld surface detection models. Furthermore, the impressive applicability of the model is verified across four scaled versions of Yolo-v5. Based on the proposed strategies, the FPS increases by more than 30 frames in the fast s and n versions of Yolo-v5. These results demonstrate the great potential of the model for industrial applications.

Ultra-wideband (UWB) radio frequency positioning technology has found extensive applications in indoor and outdoor localization due to its strong anti-interference capabilities, high penetration power, and precise measurement accuracy. However, its positioning accuracy significantly decreases under non-line-of-sight (NLOS) conditions, particularly in dynamic NLOS environments. Therefore, it is essential to identify NLOS propagation and mitigate its associated errors. To overcome these issues, this study proposes a novel enhanced Domain-Adversarial neural network for UWB signal occlusion recognition (EDANN-SOR). EDANN-SOR integrates hybrid manually extracted channel impulse response (CIR) features, effectively aligning the source and target domains and extracting discriminative and domain-invariant UWB CIR features in dynamic NLOS environments. Through adversarial learning, our method achieves high generalization with minimal CIR sample requirements, reducing the need for extensive data collection and significantly enhancing robustness in distance measurement and positioning under dynamic NLOS conditions. In contrast to state-of-the-art NLOS identification and transfer learning (TL) techniques, our method excels in both binary and multi-class NLOS classification. Specifically, it achieves an accuracy of over 97.36% in conventional LOS/NLOS binary classification across scenarios and 97.07% in multi-class classification. The integration of manually extracted features has been proven effective in improving the EDANN-SOR model’s ability to distinguish among LOS, human, and glass obstacles.