Repeatability accuracy is a key performance metric for robotic arms. To address limitations in existing monocular vision-based measurement methods, this study proposes a likelihood-based pose estimation approach. Our method first obtains initial pose estimates through optimized likelihood estimation, then iteratively refines depth information. By modeling the statistical characteristics of multiple observed poses, we derive a global theoretical pose. Within this framework, two-dimensional feature points are backprojected into three-dimensional space to form an observed point cloud. The error between this observed cloud and the theoretical feature point cloud is computed using the Iterative Closest Point (ICP) algorithm, enabling accurate quantification of repeatability accuracy. Based on 30 repeated trials at each of five target poses, the proposed method achieved repeatability positioning accuracy of 0.0115 mm, 0.0121 mm, 0.0068 mm, 0.0162 mm, and 0.0175 mm at the five poses, respectively, with a mean value of 0.0128 mm and a standard deviation of 0.0038 mm across the poses. Compared with two existing monocular vision-based methods, it demonstrates superior accuracy and stability, achieving average accuracy improvements of 0.79 mm and 1.06 mm, respectively, and reducing the standard deviation by over 85%.

Unmanned aerial vehicle (UAV) vision systems have been widely deployed for aerial monitoring applications, yet small-target detection in UAV imagery remains a significant challenge due to minimal pixel representation, substantial scale variations, complex background interference, and varying illumination conditions. Existing object detection algorithms struggle to maintain high accuracy when processing small targets with fewer than 32 × 32 pixels in UAV-captured scenes, particularly in complex environments where target-background confusion is prevalent. To address these limitations, this study proposes RMH-YOLO, a refined multi-scale architecture. The model incorporates four key innovations: a Refined Feature Module (RFM) that fuses channel and spatial attention mechanisms to enhance weak feature representation of small targets while maintaining contextual integrity; a Multi-scale Focus-and-Diffuse (MFFD) network that employs a focus-diffuse transmission pathway to preserve fine-grained spatial details from high-resolution layers and propagate them to semantic features; an efficient CS-Head detection architecture that utilizes parameter-sharing convolution to enable efficient processing on embedded platforms; and an optimized loss function combining Normalized Wasserstein Distance (NWD) with InnerCIoU to improve localization accuracy for small targets. Experimental validation on the VisDrone2019 dataset demonstrates that RMH-YOLO achieves a precision and recall of 53.0% and 40.4%, representing improvements of 8.8% and 7.4% over the YOLOv8n baseline. The proposed method attains mAP50 and mAP50:95 of 42.4% and 25.7%, corresponding to enhancements of 9.2% and 6.4%, respectively, while maintaining computational efficiency with only 1.3 M parameters and 16.7 G FLOPs. Experimental results confirm that RMH-YOLO effectively improves small-target detection accuracy while maintaining computational efficiency, demonstrating its broad application potential in diverse UAV aerial monitoring scenarios.

Photoplethysmography (PPG)-based heart rate variability (HRV) offers a cost-effective alternative to electrocardiography (ECG) for autonomic monitoring in wearable devices. We optimized signal processing on a 16-bit microcontroller by comparing 4th-order equivalent Butterworth and Elliptic IIR bandpass filters (0.8-20 Hz, zero-phase) at 1000, 500, and 250 Hz. Paired PPG-ECG recordings from 10 healthy adults were analyzed for ln HF, ln LF, and ln VLF using Lin's concordance correlation coefficient (CCC), ±5% equivalence testing (TOST), and Passing-Bablok regression (PBR). Butterworth at 500 Hz preserved near-identity with ECG standard (CCC ≥0.94; TOST met equivalence; PBR slopes/intercepts: ln HF = 0.97x + 0.10, ln LF = 1.02x - 0.07, ln VLF = 1.01x - 0.03), while halving computational load. In contrast, Elliptic at 250 Hz degraded concordance (CCC ≈ 0.64) and failed equivalence, with greater bias from nonlinear phase and ripple-induced distortion. Elliptic performance improved at higher sampling but offered no benefit over Butterworth. These results support zero-phase Butterworth filtering at ≥500 Hz as the optimal balance of fidelity, robustness, and efficiency, enabling reliable PPG-HRV monitoring on low-power devices. As a pilot investigation ( = 10), this study establishes preliminary design parameters and optimal configurations to guide subsequent large-scale clinical validation.

Accurately forecasting traffic congestion on urban expressways remains challenging, especially under unstable flow conditions where conventional machine learning models often suffer from reduced accuracy and interpretability. This study introduces a domain-theoretic machine learning framework designed for real-time congestion prediction on the Chalong Rat Expressway in Bangkok, Thailand. Feature engineering incorporates principles from the macroscopic cell transmission model, Kerner's three-phase theory, and Helbing's microscopic dynamics to capture key interactions such as density-flow relationships, jam propagation, and driver response gradients. A hybrid random forest-XGBoost ensemble is developed and evaluated against standard machine learning baselines. The results demonstrate that the proposed ensemble achieved superior performance across mean absolute error (MAE), root mean square error (RMSE), coefficient of determination (R), and prediction interval coverage (PICP), particularly near congestion transition boundaries. SHapley Additive exPlanations (SHAP) analysis confirmed corrected outflow, jam speed, and repulsive force as dominant predictors, underscoring the model's interpretability. By integrating traffic theory with interpretable machine learning, this framework enables accurate, explainable, and deployable real-time congestion forecasting for intelligent transportation systems.

The phase-sensitive optical time-domain reflectometer (Φ-OTDR) system is a distributed optical fiber sensing technology capable of measuring weak vibration signals in real time. However, while the use of a narrow-linewidth laser source enhances the system's sensitivity, the accompanying high coherence introduces an inherent drawback: fading noise. This phenomenon can lead to significant phase demodulation distortion, severely compromising the system's reliability. Consequently, interference fading represents a fundamental challenge in Φ-OTDR systems. We propose an optimized density clustering algorithm, termed adaptive principal component analysis DBSCAN++ (AP-DBSCAN). The procedure begins by identifying fading regions based on the fading principle. Subsequently, AP-DBSCAN integrates the K-distance to adaptively determine parameters, and incorporates PCA technology and the DBSCAN++ algorithm to efficiently and accurately distinguish fading points within these regions. Finally, the compromised data points are reconstructed using a nearest-neighbor interpolation method. Experimental results demonstrate the superior performance of the proposed method over DBSCAN, FDBSCAN, and DBSCAN++. Our approach achieves adaptive determination of the eps and Minpts parameters, maintaining a high fading-point detection accuracy of 99.92% while significantly improving computational efficiency by 67.33% to 76.29%.

The Editorial Office and Editorial Board of are jointly issuing a resolution and removal of the linked to this article [...].

High-voltage (HV) cable systems-particularly those insulated with cross-linked polyethylene (XLPE)-are increasingly deployed in both AC and DC applications due to their excellent electrical and mechanical performance. However, their long-term reliability is challenged by partial discharges (PD), insulation aging, space charge accumulation, and thermal and electrical stresses. This review provides a comprehensive survey of the state-of-the-art technologies and methodologies across several domains critical to the assessment and enhancement of cable reliability. It covers advanced condition monitoring (CM) techniques, including sensor-based PD detection, signal acquisition, and denoising methods. Aging mechanisms under various stressors and lifetime estimation approaches are analyzed, along with fault detection and localization strategies using time-domain, frequency-domain, and hybrid methods. Physics-based and data-driven models for PD behavior and space charge dynamics are discussed, particularly under DC conditions. The article also reviews the application of numerical tools such as FEM for thermal and field stress analysis. A dedicated focus is given to machine learning (ML) and deep learning (DL) models for fault classification and predictive maintenance. Furthermore, standards, testing protocols, and practical issues in sensor deployment and calibration are summarized. The review concludes by evaluating intelligent maintenance approaches-including condition-based and predictive strategies-framed within real-world asset management contexts. The paper aims to bridge theoretical developments with field-level implementation challenges, offering a roadmap for future research and practical deployment in resilient and smart power grids. This review highlights a clear gap in fully integrated AC/DC diagnostic and aging analyses for XLPE cables. We emphasize the need for unified physics-based and ML-driven frameworks to address HVDC space-charge effects and multi-stress degradation. These insights provide concise guidance for advancing reliable and scalable cable assessment.

Fiber optic vibration identification has significant applications in engineering fields, like security surveillance and structural health assessment. However, present methods primarily depend on either temporal-frequency domain or image features simply, challenging the simultaneous consideration of both image attributes and the temporal dependencies of vibration signals. Consequently, the performance of fiber optic vibration recognition remains subject to improvement, and its effectiveness further diminishes under conditions of uneven data distribution. Therefore, this study integrates residual neural networks, long short-term memory networks, and diffusion denoising probabilistic models to propose a fiber optic vibration recognition method DR-LSTM, which incorporates both image and temporal features while ensuring high recognition accuracy across balanced and imbalanced data distributions. Firstly, features of the Mel spectrum image and temporal characteristics of fiber optic vibration events are extracted. Subsequently, specialized neural network models are developed for categories with scarce data to produce similar images for data augmentation. Finally, the retrieved composite characteristics are employed to train recognition models, thereby improving recognition accuracy. Experiments were performed on datasets from natural environment and anthropogenic vibration, including for both balanced and imbalanced data distributions. The results show that on the two balanced datasets, the proposed model achieves improvements in classification accuracy of at least 0.67% and 7.4% compared to conventional methods. In the two imbalanced datasets, the model's accuracy exceeds that of conventional models by a minimum of 18.79% and 2.4%. This validates the effectiveness and feasibility of DR-LSTM in enhancing recognition accuracy and addressing issues with imbalanced data distribution.

Single-image dehazing suffers from severe information loss and the under-constraint problem. The lack of high-quality robust priors leads to limited generalization ability of existing dehazing methods in real-world scenarios. To tackle this challenge, we propose a simple but effective single-image dehazing network by fusing high-quality semantic priors extracted from Segment Anything Model 2 (SAM2) with different types of advanced convolutions, abbreviated SAM2-Dehaze, which follows the U-Net architecture and consists of five stages. Specifically, we first employ the superior semantic perception and cross-domain generalization capabilities of SAM2 to generate accurate structural semantic masks. Then, a dual-branch Semantic Prior Fusion Block is designed to enable deep collaboration between the structural semantic masks and hazy image features at each stage of the U-Net. Furthermore, to avoid the drawbacks of feature redundancy and neglect of high-frequency information in traditional convolution, we have designed a novel parallel detail-enhanced and compression convolution that combines the advantages of standard convolution, difference convolution, and reconstruction convolution to replace the traditional convolution at each stage of the U-Net. Finally, a Semantic Alignment Block is incorporated into the post-processing phase to ensure semantic consistency and visual naturalness in the final dehazed result. Extensive quantitative and qualitative experiments demonstrate that SAM2-Dehaze outperforms existing dehazing methods on several synthetic and real-world foggy-image benchmarks, and exhibits excellent generalization ability.

Accurate depth estimation is fundamental for underwater applications such as robotics and marine exploration. However, underwater imaging suffers from severe degradation due to light attenuation, scattering, and geometric distortion, which is compounded by the scarcity of real stereo data. To address these challenges, we propose Joint Dual-Branch Denoising (JDBD), which is a plug-in framework embedded within dual-branch depth estimation networks. JDBD performs task-aware denoising via bidirectional refinement between a monocular and a stereo pathway: the monocular branch combines Adaptive White Balance and a Red Inverse Channel Prior for color correction and haze suppression, while the stereo branch applies Joint Bilateral Filtering to reduce scattering and preserve edges. Trained on the synthetic UWStereo dataset and evaluated on the real-world SQUID dataset as well as a subset of UWStereo, JDBD achieves high depth estimation accuracy and visual fidelity in underwater scenes, demonstrating robust and adaptable performance across diverse conditions.

This study presents the development and validation of a low-cost, open-source sensor system for cyclists that automatically detects vehicle overtaking events while simultaneously monitoring air quality. The system integrates multiple ultrasonic sensors for autonomous overtaking detection and distance measurement with environmental sensors that record particulate matter, temperature, humidity, and GPS position. By combining these data streams, the system enables the analysis of correlations between traffic interactions and variations in particulate matter exposure under real-world cycling conditions. Test rides conducted in urban environments demonstrated that the system reliably identifies overtaking maneuvers and records corresponding environmental parameters. Elevated concentrations of particulate matter were observed during close vehicle passes and at traffic lights, highlighting moments of increased exposure to exhaust emissions. The automated detection mechanism eliminates the need for manual activation, ensuring complete and unbiased data collection. The modular design and energy-efficient operation of the system allow for flexible deployment in both mobile and stationary configurations. With its ability to objectively capture and relate safety and environmental data, the presented platform provides a foundation for large-scale field studies aimed at improving cyclist safety and understanding pollution exposure in urban traffic.

The accurate identification of crop pests and diseases is critical for global food security, yet the development of robust deep learning models is hindered by the limitations of existing datasets. To address this gap, we introduce DLCPD-25, a new large-scale, diverse, and publicly available benchmark dataset. We constructed DLCPD-25 by integrating 221,943 images from both online sources and extensive field collections, covering 23 crop types and 203 distinct classes of pests, diseases, and healthy states. A key feature of this dataset is its realistic complexity, including images from uncontrolled field environments and a natural long-tail class distribution, which contrasts with many existing datasets collected under controlled conditions. To validate its utility, we pre-trained several state-of-the-art self-supervised learning models (MAE, SimCLR v2, MoCo v3) on DLCPD-25. The learned representations, evaluated via linear probing, demonstrated strong performance, with the SimCLR v2 framework achieving a top accuracy of 72.1% and an F1 score (Macro F1) of 71.3% on a downstream classification task. Our results confirm that DLCPD-25 provides a valuable and challenging resource that can effectively support the training of generalizable models, paving the way for the development of comprehensive, real-world agricultural diagnostic systems.

Parkinson's disease is a neurodegenerative disorder disease that worsens over time and involves the deterioration of nerve cells in the brain. Gait analysis has emerged as a promising tool for early detection and monitoring of Parkinson's disease. However, the accurate classification of Parkinsonian gait is often compromised by missing body keypoints, particularly in critical regions like the hip and legs that are important for motion analysis. In this study, we propose RecovGait, a novel method that combines a gated initialization technique with unscented tracking to recover missing human body keypoints. The gated initialization provides initial estimates, which are subsequently refined through unscented tracking to enhance reconstruction accuracy. Our findings show that missing keypoints in the hips and legs significantly affect the classification result, with accuracy dropping from 0.8043 to 0.5217 in these areas. By using the gated initialization with an unscented tracking method to recover these occluded keypoints, we achieve an MAPE value as low as 0.4082. This study highlights the impact of hip and leg keypoints on Parkinson's disease gait classification and presents a robust solution for mitigating the challenges posed by occlusions in real-world scenarios.

This study presents the design and implementation of a surface plasmon resonance (SPR) sensor in the Kretschmann configuration, employing a gold film deposited on a flexible polydimethylsiloxane (PDMS) substrate as the SPR chip. The refractive-index sensitivity of the SPR sensor was evaluated with sodium chloride solutions of varying concentrations. Optimizing for both sensitivity and detection accuracy, the incident angle was fixed at 13°. The sensor exhibited a sensitivity of 3385.5 nm/RIU. Remarkably, the sensitivity variation was merely 1% after subjecting the sensor chip to 50 bending cycles in both forward and reverse directions. The sensor's efficacy was further validated through the detection of alcohol content in three different Chinese Baijiu samples, yielding a maximum relative error of 4.04% and a minimum error of 0.17%. Additionally, the sensor was utilized to study the adsorption behavior of glutathione (GSH) on the gold film under varying pH conditions. The findings revealed optimal immediate adsorption at pH = 12, attributed to the complete deprotonation of mercapto groups, facilitating the formation of Au-S bonds with gold atoms. The best film-forming effect was observed at pH = 7, where the interplay of attractive and repulsive forces among different molecular groups led to the gradual extension of the molecular chain, resulting in a thicker molecular film.

Social robots in public cultural venues, such as science museums, must engage diverse visitors through brief, one-off encounters where long-term user modeling is infeasible. This research examines immediately interpretable behavioral cues of a robot that can evoke a sense of personalization without storing or profiling individual users. First, a video-based observational study of expert and novice museum docents identified service strategies that enable socially adaptive communication. Building on these insights, three controlled laboratory studies investigated how specific cues from robots influence user perception. A video-based controlled study examined how recognition accuracy shapes users' social impressions of the robot's intelligence. Additional studies based on the Wizard-of-Oz (WoZ) method tested whether explanatory content aligned with participants' background knowledge and whether explicit preference inquiry and memory-based continuity strengthened perceptions of personalization. Results showed that recognition accuracy improved social impressions, whereas knowledge alignment, explicit preference inquiry, and memory-based continuity cues increased perceived personalization. These findings demonstrate that micro-level personalization cues, interpretable within a short-term encounter, can support user-centered interaction design for social robots in public environments.

Aiming at the signal interference and feature recognition difficulties existing in the detection of concealed defects such as cracks and voids behind the tunnel lining, this study carried out a 1:1 reinforced concrete-steel arch frame composite lining model test; simulated the surrounding rock defects scenarios of three types of filling media, namely crushed stone, air, and water; and analyzed the time-domain, frequency-domain, and time-frequency-domain characteristics of the geological radar signal data. The research finds that the water-filled area generates a strong reflection due to the high dielectric constant, with the spectral peak reaching 712 MHz and the high-frequency component significantly enhanced. The peak frequency of the air-filled zone spectrum is 531 MHz, and the high-frequency bandwidth is broadened. The spectral peak of the crushed stone filling area is 507 MHz, with fast high-frequency attenuation and energy dispersion. The time-domain waveforms show that the amplitude in the water-filled area is the highest and the tailing is obvious, the waveform in the air-filled area is sharp, and the amplitude in the crushed stone-filled area is gentle. The peak frequency of the spectrum, the amplitude attenuation law, and the waveform shape can be used as the key indicators for discriminating the category of filling materials. The analysis method of feature fusion in the time-frequency domain has important engineering application value for improving the detection accuracy of geological radar in complex lining structures.

Deep neural networks have outperformed conventional methods in various fields such as image recognition, natural language processing, and speech recognition. In particular, vision models are widely applied to real-world domains including medical image analysis, autonomous driving, smart factories, and security surveillance. However, these models are vulnerable to adversarial attacks, which pose serious threats to safety and reliability. Among different attack types, this study focuses on evasion attacks that perturb the inputs of deployed models, with an emphasis on black-box settings. The zeroth order optimization (ZOO) attack can approximate gradients and execute attacks without access to internal model information, but it becomes inefficient and exhibits low success rates on high-resolution images due to its dependence on image resizing and its high memory complexity. To address these limitations, this study proposes a patch-based fast zeroth order optimization attack, FP-ZOO. FP-ZOO partitions images into patches and generates perturbations effectively by employing probability-based sampling and an ϵ-greedy scheduling strategy. We conducted a large-scale evaluation of the FP-ZOO attack on the CIFAR-10, CIFAR-100, and ImageNet datasets. In this evaluation, we adopted attack success rate, L2 distance, and adversarial example generation time as performance metrics. The evaluation results showed that the FP-ZOO attack not only achieved an attack success rate of 97-100% against ImageNet in untargeted attacks, but also demonstrated performance up to 10 s faster compared to the ZOO attack. However, in targeted attacks, it showed relatively lower performance compared to baseline attacks, leaving it as a future research topic.

Autonomous vehicles represent a transformative technology in modern transportation, promising enhanced safety, efficiency, and accessibility in mobility systems. This paper presents a comprehensive autonomous vehicle system designed specifically for Vietnam's traffic conditions, featuring a multi-layered approach to perception, decision-making, and control. The system utilizes dual 2D LiDARs, camera vision, and GPS sensing to navigate complex urban environments. A key contribution is the development of a specialized segmentation model that accurately identifies Vietnam-specific traffic signs, lane markings, road features, and pedestrians. The system implements a hierarchical decision-making architecture, combining long-term planning based on GPS and map data with short-term reactive planning derived from a bird's-eye view transformation of segmentation and LiDAR data. The control system modulates the speed and steering angle through a validated model that ensures stable vehicle operation across various traffic scenarios. Experimental results demonstrate the system's effectiveness in real-world conditions, achieving a high accuracy rate in terms of segmentation and detection and an exact response in navigation tasks. The proposed system shows robust performance in Vietnam's unique traffic environment, addressing challenges such as mixed traffic flow and country-specific road infrastructure.

Predictive maintenance is increasingly important in rotating machinery to prevent unexpected failures, reduce downtime, and improve operational efficiency. This study compares the efficacy of traditional machine learning (ML) and deep learning (DL) techniques in diagnosing bearing faults under varying load and speed conditions. Two classification tasks were conducted: a simpler three-class task that distinguishes healthy bearings, inner race faults, and outer race faults, and a more complex nine-class task that includes faults of varying severity in the inner and outer races. In this study, the machine learning algorithm ensemble bagged trees, achieved maximum accuracies of 93.04% for the three-class and 87.13% for the nine-class classifications, followed by neural network, SVM, KNN, decision tree, and other algorithms. For deep learning, the CNN model, trained on scalograms (time-frequency images generated by continuous wavelet transform), demonstrated superior performance, reaching up to 100% accuracy in both classification tasks after six training epochs for the nine-class classifications. While CNNs take longer training time, their superior accuracy and capability to automatically extract complex features make the investment worthwhile. Consequently, the results demonstrate that the CNN model trained on CWT-based scalogram images achieved remarkably high classification accuracy, confirming that deep learning methods can outperform traditional ML algorithms in handling complex, non-linear, and dynamic diagnostic scenarios.

Enhancing sustainability in agriculture has become a significant challenge today where in the current context of climate change, particularly in countries of the Mediterranean area, the amount of water available for irrigation is becoming increasingly limited. Automating irrigation processes using affordable sensors can help save irrigation water and produce almonds more sustainably. This work presents an IoT-enabled edge computing model for smart irrigation systems focused on precision agriculture. This model combines IoT sensors, hybrid machine learning algorithms, and edge computing to predict soil moisture and manage Controlled Deficit Irrigation (CDI) strategies in high density almond tree fields applying reductions of 35% ETc (crop evapotranspiration). By gathering and analyzing meteorological, humidity soil, and crop data, a soft ML (Machine Learning) model has been developed to enhance irrigation practices and identify crop anomalies in real-time without cloud computing. This methodology has the potential to transform agricultural practices by enabling precise and efficient water management, even in remote locations with lack of internet access. This study represents an initial step toward implementing ML algorithms for irrigation CDI strategies.

In the original publication [...].