Search Results (12,932)

Labor shortages and uneven terrain in orchards present significant challenges to autonomous navigation. This study proposes a navigation system that integrates Light Detection and Ranging (LiDAR) and Inertial Measurement Unit (IMU) data to enhance localization accuracy and map stability through Simultaneous Localization and Mapping (SLAM). To minimize distortions in LiDAR scans caused by ground irregularities, real-time tilt correction was implemented based on IMU feedback. Furthermore, the path planning module was improved by modifying the Rapidly-Exploring Random Tree (RRT) algorithm. The enhanced RRT generated smoother and more efficient trajectories with quantifiable improvements: the average shortest path length was 2.26 m, compared to 2.59 m with conventional RRT and 2.71 m with A* algorithm. Tracking performance also improved, achieving a root mean square error of 0.890 m and a maximum lateral deviation of 0.423 m. In addition, yaw stability was strengthened, as heading fluctuations decreased by approximately 7% relative to the standard RRT. Field results validated the robustness and adaptability of the proposed system under real-world agricultural conditions. These findings highlight the potential of LiDAR–IMU sensor fusion and optimized path planning to enable scalable and reliable autonomous navigation for precision agriculture. Full article

Fault detection and classification localization in photovoltaic power grids is a key challenge in photovoltaic power systems. Due to the greater fluctuation of power data in photovoltaic power grids, traditional grid fault detection methods suffer from inefficiency, low accuracy, and inaccurate fault localization in photovoltaic scenarios. In this paper, a fuzzy control technique combined with an improved GABP neural network is used to identify potential fault nodes in the photovoltaic distribution network. The symmetric crossover operator of the genetic algorithm and the symmetry constraints of the neural network weight matrix are used to improve the model’s ability to capture the symmetric fluctuation characteristics of photovoltaic data, while a classification module consisting of three fuzzy controllers is used for fault identification. The simulation results show that the recognition method proposed in this paper has good performance and the fault classification accuracy reaches 92.75%, which provides a practical reference value for the management of photovoltaic distribution network. Full article

This paper investigates modulation recognition technology for high-order modulated signals. Addressing the issue that existing deep learning-based modulation recognition methods struggle to effectively capture the features of long sequence signals in high-order modulation, we propose a ConvMamba model that integrates convolutional neural networks (CNNs) with the Mamba2 architecture. By employing a selective state-space model, the ConvMamba effectively captures the temporal dependencies in long sequence signals. It also combines the local feature extraction capability of CNNs with a soft-thresholding denoising module, forming a hybrid structure that possesses both global modeling and noise resistance capabilities. The evaluation results on the Sig53 dataset, which contains a rich variety of high-order modulations, demonstrate that compared to traditional CNN- or Transformer-based architectures, ConvMamba achieves a better balance between computational efficiency and recognition accuracy. Compared to Transformer models with similar performance, ConvMamba reduces computational complexity by over 60%. Compared to CNN models with comparable computational resource consumption, ConvMamba significantly improves recognition accuracy. Therefore, ConvMamba shows a distinct advantage in processing high-order modulated signals with long sequences. Full article

Multi-temporal remote sensing image matching plays a crucial role in tasks such as detecting changes in urban buildings, monitoring agriculture, and assessing ecological dynamics. Due to temporal variations in images, significant changes in land features can lead to low accuracy or even failure when matching results. To address these challenges, in this study, a remote sensing image matching framework is proposed based on multi-perception and enhanced feature description. Specifically, the framework consists of two core components: a feature extraction network that integrates multiple perceptions and a feature descriptor enhancement module. The designed feature extraction network effectively focuses on key regions while leveraging depthwise separable convolutions to capture local features at different scales, thereby improving the detection capabilities of feature points. Furthermore, the feature descriptor enhancement module optimizes feature point descriptors through self-enhancement and cross-enhancement phases. The enhanced descriptors not only extract the geometric information of the feature points but also integrate global contextual information. Experimental results demonstrate that, compared to existing remote sensing image matching methods, our approach maintains a strong matching performance under conditions of angular and scale variation. Full article

Paratuberculosis is a widespread infectious disease in ruminants that leads to significant economic losses in livestock production. In this study, we developed a practical method for predicting the likelihood of the herd-level presence of the infection and estimating its prevalence in subgroups of a dairy herd—specifically, first-time calving cows (primiparous) and those that have calved more than once (multiparous). We fit a Bayesian hierarchical model to cow-level data, incorporating prior knowledge about regional prevalence of infection to improve the accuracy and reliability of the estimates. The model was tested using synthetic data representing six regional scenarios in four countries (Chile, Denmark, Italy, and Hungary). The likelihood that a herd is infected is evaluated using Bayes factors and posterior probability of infection. Both the Bayes factor and the posterior probability of infection classified the simulated herds in accordance with the proportions of infected herds. Summary measures obtained for within-herd true prevalence estimates demonstrated acceptable accuracy. The R and STAN codes of the model are available as an open-access tool. The model can be customized for any region using real local data and prior information. The relationship between true and apparent prevalence is linear and stable and therefore can be estimated well. We found that, in Hungary, the TP/AP ratios were 1.6 and 1.5 for primi- and multiparous cows, respectively. Full article

With the widespread deployment of 5G technology, high-precision positioning in global navigation satellite system (GNSS)-denied environments is a critical yet challenging task for emerging 5G applications, enabling enhanced spatial resolution, real-time data acquisition, and more accurate geolocation services. Traditional methods relying on single-source measurements like received signal strength information (RSSI) or time of arrival (TOA) often fail in complex multipath conditions. To address this, the positional encoding multi-scale residual network (PE-MSRN) is proposed, a novel deep learning framework that enhances positioning accuracy by deeply mining spatial information from 5G channel state information (CSI). By designing spatial sampling with multigranular data and utilizing multi-source information in 5G CSI, a dataset covering a variety of positioning scenarios is proposed. The core of PE-MSRN is a multi-scale residual network (MSRN) augmented by a positional encoding (PE) mechanism. The positional encoding transforms raw angle of arrival (AOA) data into rich spatial features, which are then mapped into a 2D image, allowing the MSRN to effectively capture both fine-grained local patterns and large-scale spatial dependencies. Subsequently, the PE-MSRN algorithm that integrates ResNet residual networks and multi-scale feature extraction mechanisms is designed and compared with the baseline convolutional neural network (CNN) and other comparison methods. Extensive evaluations across various simulated scenarios, including indoor autonomous driving and smart factory tool tracking, demonstrate the superiority of our approach. Notably, PE-MSRN achieves a positioning accuracy of up to 20 cm, significantly outperforming baseline CNNs and other neural network algorithms in both accuracy and convergence speed, particularly under real measurement conditions with higher SNR and fine-grained grid division. Our work provides a robust and effective solution for developing high-fidelity 5G positioning systems. Full article

Wet cracks caused by leakage often exhibit visual and structural distortions due to surface contamination, salt crystallization, and corrosion byproducts. These factors significantly degrade the performance of sensor- and vision-based crack detection systems. In moist environments, the initiation and propagation of cracks tend to be highly nonlinear and irregular, making it challenging to distinguish crack regions from the background—especially under visual noise such as reflections, stains, and low contrast. To address these challenges, this study proposes a segmentation framework that integrates a dedicated preprocessing pipeline aimed at suppressing noise and enhancing feature clarity, all without altering the underlying segmentation architecture. The pipeline begins with adaptive thresholding to perform initial binarization under varying lighting conditions. This is followed by morphological operations and connected component analysis to eliminate micro-level noise and restore structural continuity of crack patterns. Subsequently, both local and global contrast are enhanced using histogram stretching and contrast limited adaptive histogram equalization. Finally, a background fusion step is applied to emphasize crack features while preserving the original surface texture. Experimental results demonstrate that the proposed method significantly improves segmentation performance under adverse conditions. Notably, it achieves a precision of 97.5% and exhibits strong robustness against noise introduced by moisture, reflections, and surface irregularities. These findings confirm that targeted preprocessing can substantially enhance the accuracy and reliability of crack detection systems deployed in real-world infrastructure inspection scenarios. Full article

During machining, most of the mechanical energy is converted into heat. A substantial part of this heat is transferred to the cutting tool, causing a rapid rise in tool temperature. Excessive thermal loads accelerate tool wear and lead to displacement of the tool center point, reducing machining accuracy and workpiece quality. This challenge is particularly pronounced when machining titanium alloys. Due to their low thermal conductivity, titanium alloys impose significantly higher thermal loads on the cutting tool compared to conventional carbon steels, making the process more difficult. To reduce temperatures in the cutting zone, cutting fluids are widely employed in titanium machining. They have been shown to significantly extend tool life. Cutting fluids are broadly categorized into cutting oils and water-based cutting fluids. Owing to their distinct thermophysical properties, these fluids exhibit notably different cooling and lubrication performance. However, current research lacks comprehensive cross-comparative studies of different cutting fluid types, which hinders the selection of optimal cutting fluids for process optimization. This study examines the influence of three cutting fluids—emulsion, cutting oil, and synthetic oil-free fluid—on tool wear, temperature, surface quality, and energy consumption during flood-cooled end milling of Ti-6Al-4V. A novel experimental setup incorporating embedded thermocouples enabled real-time temperature measurement near the cutting edge. Tool wear, torque, and surface roughness were recorded over defined feed lengths. Among the tested fluids, emulsion achieved the best balance of cooling and lubrication, resulting in the longest tool life with a feed travel path of 12.21 m. This corresponds to an increase of approximately 200 % compared to cutting oil and oil-free fluid. Cutting oil offered superior lubrication but limited cooling capacity, resulting in localized thermal damage and edge chipping. Water-based cutting fluids reduced tool temperatures by over 300 ${}^{\circ }$C compared to dry cutting but, in some cases, increased notch wear due to higher mechanical stress at the entry point. Power consumption analysis revealed that the cutting fluid supply system accounted for 60–70 % of total energy use, particularly with high-viscosity fluids like cutting oil. Complementary thermal and CFD simulations were used to quantify heat partitioning and convective cooling efficiency. The results showed that water-based fluids achieved heat transfer coefficients up to 175 kW/m²· K, more than ten times higher than those of cutting oil. These findings emphasize the importance of selecting suitable cutting fluids and optimizing their supply to enhance tool performance and energy efficiency in Ti-6Al-4V machining. Full article

Chatter vibration in machining operations has been identified as one of the major obstacles to improving surface quality and productivity. Therefore, efficiently and accurately predicting stable cutting regions is becoming increasingly important, especially in high-speed milling processes. In this study, on the basis of a predictor–corrector scheme, the following three error correction methods are developed for milling stability analysis: the Correction Hamming–Milne-based method (CHM), the Correction Adams–Milne-based method (CAM) and the Predictor–Corrector Hamming–Adams–Milne-based method (PCHAM). Firstly, we employ the periodic delay differential equations (DDEs), which are usually adopted to describe mathematical models of milling dynamics, and the time period of the coefficient matrix is divided into two unequal subintervals based on an analysis of the vibration modes. Then, the Hamming method and the fourth-order implicit Adams–Moulton method are separately utilized to predict the state term, and the Milne method is adopted to correct the state term. Based on local truncation error, combining the Hamming and Milne methods creates a CHM that can more precisely approximate the state term. Similarly, combining the fourth-order implicit Adams–Moulton method and the Milne method creates a CAM that can more accurately approximate the state term. More importantly, the CHM and the CAM are employed together to acquire the state transition matrix. Thereafter, the effectiveness and applicability of the three error correction methods are verified by comparing them with three existing methods. The results demonstrate that the three error correction methods achieve higher prediction accuracy without sacrificing computational efficiency. Compared with the 2nd SDM, the calculation times of the CHM, CAM and PCHAM are reduced by around 56%, 56% and 58%, respectively. Finally, verification experiments are carried out using a CNC machine (EMV650) to further validate the reliability of the proposed methods, where ten groups of cutting tests illustrate that the stability lobes predicted by the three error correction methods exhibit better agreement with the experimental results. Full article

Fish diseases are one of the primary causes of economic losses in aquaculture. Existing deep learning models have progressed in fish disease detection and lesion segmentation. However, many models still have limitations, such as detecting only a single type of fish disease or completing only a single task within fish disease detection. To address these limitations, we propose FDMNet, a multi-task learning network. Built upon the YOLOv8 framework, the network incorporates a semantic segmentation branch with a multi-scale perception mechanism. FDMNet performs detection and segmentation simultaneously. The detection and segmentation branches use the C2DF dynamic feature fusion module to address information loss during local feature fusion across scales. Additionally, we use uncertainty-based loss weighting together with PCGrad to mitigate conflicting gradients between tasks, improving the stability and overall performance of FDMNet. On a self-built image dataset containing three common fish diseases, FDMNet achieved 97.0% mAP50 for the detection task and 85.7% mIoU for the segmentation task. Relative to the multi-task YOLO-FD baseline, FDMNet’s detection mAP50 improved by 2.5% and its segmentation mIoU by 5.4%. On the dataset constructed in this study, FDMNet achieved competitive accuracy in both detection and segmentation. These results suggest potential practical utility. Full article

A lightweight enhanced detection model named YOLO-LDFI is proposed in this study for ship target detection in SAR images, aiming to improve detection accuracy and deployment efficiency under complex maritime environments. Based on YOLOv11n, the model incorporates four architectural improvements in a progressive manner: linear deformable convolution (LDConv), deformable context-aware attention mechanism (DCAM), frequency-adaptive dilated convolution detection head (FAHead), and Inner-EIoU. Experiments conducted on the public SAR ship detection dataset HRSID demonstrate that the proposed model achieves an AP50 of 90.7% and an F1 score of 87.0%, with only 2.63 M parameters and a computational complexity of 6.7 GFLOPs. Ablation experiments validate the contribution of each component to improved feature alignment, reduced background interference, and more accurate target localization. Overall, the results indicate that the proposed model offers a reasonable trade-off between detection performance and computational efficiency in SAR ship detection tasks. Full article

Simultaneous Localization and Mapping (SLAM) systems require accurate and globally consistent mapping to ensure the long-term stable operation of robots or vehicles. However, for the commercial applications of indoor sweeping robots, the system needs to maintain accuracy while keeping computational and storage requirements low to ensure cost controllability. This paper proposes a dual-point-cloud-registration SLAM based on line features for the mapping of a mobile robot, named DPCR-SLAM. The front-end employs an improved Point-to-Line Iterative Closest Point (PLICP) algorithm for point cloud registration. It first aligns the point cloud and updates the submap. Subsequently, the submap is aligned with the regional map, which is then updated accordingly. The back-end uses the association between regional maps to perform graph optimization and update the global map. The experimental results show that, in the application scenario of indoor sweeping robots, the proposed method reduces the map storage space by 76.3%, the point cloud processing time by 55.8%, the graph optimization time by 77.7%, and the average localization error by 10.9% compared to the Cartographer, which is commonly used in the industry. Full article

Spatial heterogeneity in soil deposition poses a significant challenge to accurate geotechnical characterization, which is essential for sustainable infrastructure development. This study presents an advanced geotechnical data-driven mapping framework, based on a monotonized and augmented formulation of Shepard’s inverse distance weighting (IDW) algorithm, implemented through the Google Earth Engine (GEE) platform. The approach is rigorously evaluated through a comparative analysis against the classical IDW and Kriging techniques using standard key performance indices (KPIs). Comprehensive field and laboratory data repositories were developed in accordance with international geotechnical standards (e.g., ASTM). Key geotechnical parameters, i.e., standard penetration test (SPT-N) values, shear wave velocity (Vs), soil classification, and plasticity index (PI), were used to generate high-resolution geospatial models for a previously unmapped region, thereby providing essential baseline data for building infrastructure design. The results indicate that the augmented IDW approach exhibits the best spatial gradient conservation and local anomaly detection performance, in alignment with Tobler’s First Law of Geography, and outperforms Kriging and classical IDW in terms of predictive accuracy and geologic plausibility. Compared to classical IDW and Kriging, the augmented IDW algorithm achieved up to a 44% average reduction in the RMSE and MAE, along with an approximately 30% improvement in NSE and PC. The difference in spatial areal coverage was found to be up to 20%, demonstrating an improved capacity to model spatial subsurface heterogeneity. Thematic design maps of the load intensity (LI), safe bearing capacity (SBC), and optimum foundation depth (OD) were constructed for ready application in practical design. This work not only establishes the inadequacy of conventional geostatistical methods in highly heterogeneous soil environments but also provides a scalable framework for geotechnical mapping with accuracy in data-poor environments. Full article

In response to problems such as large target scale variations, strong background noise, and blurred features leading by low contrast in infrared target detection in near space environments, this paper proposes an efficient detection model, YOLO-MARS, which is based on YOLOv8. The model introduces a Space-to-Depth (SPD) convolution module into the backbone section, which retains the detailed features of smaller targets by downsampling operations without information loss, alleviating the loss of the target feature caused by traditional downsampling. The Grouped Multi-Head Self-Attention (GMHSA) module is added after the backbone’s SPPF module to improve cross-scale global modeling capabilities for target area feature responses while suppressing complex thermal noise background interference. In addition, a Light Adaptive Spatial Feature Fusion (LASFF) detector head is designed to mitigate the scale sensitivity issue of infrared targets (especially smaller targets) in the feature pyramid. It uses a shared weighting mechanism to achieve adaptive fusion of multi-scale features, reducing computational complexity while improving target localization and classification accuracy. To address the extreme scarcity of near space data, we integrated 284 near space images with the HIT-UAV dataset through physical equivalence analysis (atmospheric transmittance, contrast, and signal-to-noise ratio) to construct the NS-HIT dataset. The experimental results show that ${mAP}_{@0.5}$ increases by 5.4% and the number of parameters only increase 10% using YOLO-MARS compared to YOLOv8. YOLO-MARS improves the accuracy of detection significantly while considering the requirements of model complexity, which provides an efficient and reliable solution for applications in near space infrared target detection. Full article

To address the challenges of dynamic coupling interference and time-frequency feature degradation in current approaches to Ultra-High-Frequency Radio-Frequency Identification (UHF RFID) behavior recognition, this study proposes a novel behavior recognition method integrating multi-feature analysis with a dual-path residual network. The proposed method mitigates interference by using phase difference methods to eliminate signal errors and cross-correlation, as well as adaptive equalization algorithms to decouple interfering signals. To identify the target tags participating in behavioral interactions, we construct a three-dimensional feature space and apply an improved weighted isolated forest algorithm to detect active tags during interactions. Subsequently, Doppler shift analysis extracts behavioral features, and multiscale wavelet-packet decomposition generates time-frequency representations. The dual-path residual network then fuses global and local features from these time-frequency representations for behavioral classification, thereby identifying interaction behaviors such as ‘taking away’, ‘putting back’, and ‘hesitation’ (characterized by picking up, then putting back). Experimental results demonstrate that the proposed scheme achieves behavioral recognition accuracy of 94% in complex scenarios, significantly enhancing the overall robustness of interaction behavior recognition. Full article