Sensors doi: 10.3390/s26113452

Authors: Yongjie Wang Conghui Zhu Wenxi Zhang

Point diffraction interferometers (PDIs) utilize a near-ideal spherical wavefront generated by point diffraction as the reference, providing a high-quality measurement benchmark independent of reference surface quality. In this work, a compact dual-oblique-fiber heterodyne phase-shifting point diffraction interferometer (DOF-HPSPDI) is proposed. A dual-oblique-fiber point diffraction wavefront generator (DOF-PDWG) is designed to generate the reference and measurement beams separately. The proposed configuration enables efficient utilization of the divergence of the fiber-generated diffracted wavefront, while the reflective structure at the fiber end faces allows the two beams to propagate along a common path. In addition, the close spacing between the two oblique fibers minimizes system errors. Heterodyne phase-shifting interferometry (HPSI) is employed to retrieve the wavefront phase from the interferograms. Theoretical system errors are analyzed through simulations, and experiments verify the feasibility and stability of the proposed system. This work provides a low-cost, compact, and highly stable point diffraction interferometer, offering a promising device for high-precision optical testing and sub-aperture stitching of large-aperture optical components.

Sensors doi: 10.3390/s26113451

Authors: Dennis Birkenmaier Shanthan Rao Kanuganti Wilhelm Stork

Accurate stress monitoring is critical for high-risk professions like firefighting, yet existing wearable solutions face challenges balancing accuracy with practical usability. While electrodermal activity (EDA) offers a non-invasive, single-sensor approach, current automated feature extraction methods fail to capture stress-discriminative patterns effectively. We developed a hybrid stress detection pipeline combining 20 hand-crafted physiological features with 32 deep-learned features from a supervised convolutional autoencoder. Unlike traditional unsupervised approaches optimized solely for signal reconstruction, our architecture employs a dual-head design with weighted classification loss to guide feature learning toward stress discrimination. The system was validated on the WESAD dataset (15 subjects) using rigorous leave-one-subject-out (LOSO) cross-validation, along with comprehensive preprocessing, including cvxEDA decomposition, adaptive artifact detection, and physiological peak validation. Our optimized K-Nearest Neighbors classifier achieved 98.62% accuracy, surpassing the industry-standard PyEDA benchmark (97.0%) by 1.62 percentage points. The model demonstrated 97.58% sensitivity (true positive rate) and 98.92% specificity (true negative rate), with only 2.42% false negatives—critical for safety-critical applications. Ablation studies revealed that unsupervised autoencoder features alone achieved only 55% accuracy, increasing to 89% with supervised learning and 98.62% with the hybrid approach, representing a 43.62-percentage-point improvement. This work demonstrates that combining domain-specific physiological knowledge with label-aware deep learning produces more discriminative features than either approach alone. The resulting system successfully translates complex probabilistic outputs into an interpretable 1–10 stress score, providing a practical foundation for real-time stress monitoring in wearable devices for first responders.

Sensors doi: 10.3390/s26113450

Authors: Baohui Zhang Haoyu Zhou Xiaowu Zhu Haoxiang Chen Yang Yang

High-performance ethanol sensors with low power consumption show critical applications in environmental monitoring, personal health diagnosis, industry and traffic safety. Herein, MoO3/Nb2C MXene heterojunction gas-sensing materials were constructed via a one-step hydrothermal method for MoO3 nanotube synthesis. The dominant facets of MoO3 were shifted from the (040) orientation in MoO3 nanotubes to the (110) and (021) orientations in the MoO3/Nb2C MXene composite. Nb2C nanosheets provide a large number of crystallization sites, preventing the growth of MoO3 nanotubes during synthesis, inducing a strategic facet release. The sensing performance shows MoO3/Nb2C MXene composite reduces the operating temperature down to 120 °C. The 15 wt% Nb2C MXene-precursor-mixed MoO3 sensor exhibits an enhanced response of 6.1 toward 100 ppm ethanol, which is higher than that of pristine MoO3 nanotubes at 120 °C, with response and recovery times of 19 s and 72 s, respectively. The sensors show high selectivity toward ethanol over other VOC gases and good long-term stability over 30 days. This work confirms that crystal engineering is an effective method for reducing operating temperature and enhancing gas-sensing performance, and the sensor shows potential application for ethanol sensing.

Sensors doi: 10.3390/s26113449

Authors: Minghao Li Ruixin Fu Dongsheng Wu Lijuan Zhao

As a core crushing equipment in mining, building materials, and related industries, the cone crusher relies heavily on the optimal design and health state prediction of its mantle liner to enhance equipment reliability and reduce maintenance costs. This paper proposes a comprehensive approach integrating dynamic modeling, intelligent optimization, and health prognosis. First, a virtual prototype model is established based on laminated crushing theory and multibody dynamics simulation to analyze the motion and force characteristics of the mantle liner. Second, for the two key parameters—counterweight mass and motor speed—an improved butterfly optimization algorithm (IBOA) incorporating Cauchy mutation and an adaptive weight is proposed to achieve efficient global optimization. Furthermore, vibration signal features are extracted at different wear stages; a comprehensive health indicator curve is constructed by combining PCA dimensionality reduction with adaptive feature fusion (ASFF), and the Weibull degradation model is employed for life extrapolation prediction. Finally, fuzzy C-means (FCM) clustering is applied to autonomously partition the health states. Parameter optimization reduces the standard deviation of the force acting on the mantle liner by approximately 15.4%, markedly improving system operational stability. Health prognosis reveals that the liner enters a faulty state after 785 h, and the health condition is effectively classified into four stages: healthy, good, degraded, and faulty. The results demonstrate that the proposed optimization and health prognosis methods can effectively improve the operational efficiency and reliability of cone crushers, exhibit favorable engineering applicability, and provide a quantitative basis for condition monitoring and maintenance decision-making.

Sensors doi: 10.3390/s26113446

Authors: Shun-Hsuan Huang Chow-Shing Shin

Adhesive joints are widely used in structural applications. However, they are susceptible to degradation under service loads and adverse environmental conditions, leading to eventual catastrophic failure. Thus, the advancement of monitoring tools that can deliver real-time data on the deterioration of adhesive joints is crucial for enhancing the reliability of structures. This study investigated the feasibility of using electrical impedance responses to monitor integrity degradation under tensile and fatigue loading in single-lap adhesive joints in aluminum alloy and carbon fiber-reinforced polymer (CFRP) specimens. Previous works on electrical impedance monitoring of adhesive joint integrity invariably employed conductive adhesives. Theoretical considerations based on the concept of a capacitive system indicate that electrical impedance monitoring may still be feasible even if the joint is non-conductive. This has important implications as it suggests that the structural health of many existing ordinary adhesive joints may be amenable to impedance-based monitoring. To test this possibility, neat epoxy adhesive joints without the addition of conductive constituents were fabricated with aluminum and composite adherends. The specimens were subjected to tensile and fatigue degradation while the impedance responses under different excitation frequencies were monitored. The results showed that impedance monitoring is insensitive for detecting damage during tensile failure because the onset of debonding that produces a detectable impedance change occurs too close to the unstable final failure. For fatigue cycling, debonding developed at an early stage and evolved in a stable manner, and the impedance gradually increased with the number of fatigue cycles, reflecting the development of fatigue damage. These findings indicate that impedance-based monitoring on non-conductive adhesive joints has strong potential for tracking structural integrity degradation, particularly for fatigue loading.

Sensors doi: 10.3390/s26113448

Authors: Zhongyu Shu Yubo Gao Guo Zhang Zihan Xu Jianping Wang

Quantifying river surface velocity and discharge is essential for flood control and mitigation. Traditional contact measurement methods are capable of providing precise results, yet they demand considerable manpower and material resources and face implementation challenges in flood seasons. Image velocimetry methods have attracted extensive attention due to their low cost, simplicity in operation, and safety. However, most of them lack a physical basis and interpretability. This paper introduces a river flow estimation algorithm combined with Physics-Informed Neural Networks (PINNs). The introduction of the convection–diffusion equation based on optical flow enables the model to better fit the flow characteristics of water and provides stronger physical support for the measurement results. The adoption of this equation as the loss function and the introduction of multiple scenarios eliminate the need for labeled data in the PINNs training process. The experimental results in both artificial and natural river channels demonstrate that the relative errors of the discharge measured by the proposed method are 0.66% and −1.75%, and the relative errors of the mean velocity are 0.64% and −2.33%. Compared with other methods, the proposed method exhibits superior performance.

Sensors doi: 10.3390/s26113447

Authors: Lan Wu Haozhe Wang Qian Li

Grain intake is a key operation in grain storage that directly affects storage efficiency, operational safety, and grain quality. In grain-entry scenarios, single LiDAR sensors are easily limited by blind spots and occlusions, making multi-LiDAR collaborative perception necessary for reliable three-dimensional environment sensing. However, heterogeneous LiDARs differ in scan lines, point density, viewing angle, installation pose, and noise characteristics, which leads to low-overlap and mixed sparse–dense point cloud registration challenges. To address this issue, this paper proposes a GICP-based registration flow improvement method for heterogeneous multi-LiDAR systems used in intelligent grain warehousing robots. The method improves registration stability through overlap-region cropping, voxel downsampling, and a star-topology registration strategy, and further introduces a point-to-plane evaluation metric based on local planar models together with cross-LiDAR planar consistency verification. Experimental results show that the proposed method reduces the point-to-plane error to 0.1487 m in the L0−L1 registration task and 0.1090 m in the L1−L2 registration task, outperforming ICP, point-to-plane ICP, and NDT while maintaining acceptable computational efficiency. These results demonstrate that the method can improve structural alignment quality and provide reliable geometric support for multi-sensor perception, mapping, and autonomous operation of grain warehousing robots. Rather than proposing a fundamentally new registration mathematical model, this study proposes a highly engineered GICP-based workflow. It should be noted that the proposed workflow is specifically tailored and optimized for plane-dominated and semi-static grain storage environments, restricting its validated scope to static or low-speed multi-LiDAR registration tasks.

Sensors doi: 10.3390/s26113445

Authors: Lirong Kou Xiaoyang Gao

With the rapid development of intelligent shipping and the autonomy of marine engineering equipment, numerous studies have focused on the advancement of Autonomous Surface Vehicles (ASVs). As a fundamental component of ASV automation systems, path planning directly determines the safety and economy of ship navigation. This paper systematically reviews recent research progress in ASV path planning. First, five key issues are identified for ASV path planning: navigation environment, environment modeling, ship motion model, collision avoidance for safety, and optimization. Second, existing algorithms are classified into four categories: graph search-based, sampling-based, numerical optimization-based, and artificial intelligence-based. The improvement directions and application scenarios of each category are elaborated. Finally, the four types of algorithms are evaluated against three indicators: path quality, scalability and extensibility, and algorithm performance. Analysis of the reviewed literature shows that traditional graph search and sampling algorithms perform well in various aspects under static environments, but are insufficient in adapting to multiple constraints and generalizing to different environments. In contrast, artificial intelligence algorithms represented by deep reinforcement learning exhibit significant advantages in dynamic collision avoidance decision-making, multi-agent coordination, and environmental generalization, and have become the mainstream direction of current research. This paper summarizes the existing challenges in safety and optimization in current ASV path planning research and prospects future development directions.

Sensors doi: 10.3390/s26113444

Authors: Yu Hao Kangning Wang Li Zhang Zibo Yuan

Combining Unmanned Aerial Vehicle (UAV) remote sensing with computer vision has become an efficient approach to detect forest wildfires. Nevertheless, existing methods still face several challenges, including missed detection of small fire spots and slender smoke under limited computational resources, false alarms caused by complex forest backgrounds, and insufficient adaptability to the irregular and dynamic morphology of fire and smoke. To address these issues, an improved YOLO26-based model, termed ASCA-YOLO, is proposed. Specifically, FWAMSConv module is introduced to improve multi-scale indicator representation of small and sparse targets. In addition, FWSCSAttention mechanism is designed to reduce background interference by modeling contextual feature distributions. Moreover, FWASIoU loss is developed to improve bounding box regression for non-rigid targets. The experimental evaluation indicates that, relative to YOLO26, the proposed model decreases the parameter count and FLOPs by 19.2% and 21.3%, respectively. Meanwhile, recall reaches 0.809 and precision reaches 0.870, indicating improved detection performance under complex conditions. In addition, mAP50-95 is improved by 12.9%, reflecting more stable localization for irregular wildfire targets. Overall, ASCA-YOLO attains a better equilibrium between detection quality and computational cost than several mainstream object detection models, indicating its potential for real-time UAV-based wildfire monitoring.

Sensors doi: 10.3390/s26113443

Authors: Yifang Tan Lijun Yuan Chuanjiang Xie Chao Zhou Xin Li Xinyu Zhu

Visible–thermal tiny pedestrian detection in UAV aerial images is crucial for online decision-making in urban security and disaster response. However, the extremely small scale and sparse distribution of pedestrians cause discriminative cues to be submerged by dominant low-frequency background and contextual redundancy during feature learning. Meanwhile, cross-modal spatial misalignment and spatially varying modality reliability hinder stable fine-grained correspondence, thereby degrading fusion quality. To address these issues, QA2FDet is proposed as a quality-aware adaptive alignment fusion network comprising three modules: spectrum-spatial decoupled enhancement module (SDE), cross-modal correspondence mining module (CCM), and prior-informed gated fusion (PGF). SDE leverages the discrete cosine transform to disentangle redundant low-frequency background information, while deep semantic gating propagates high signal-to-noise ratio details into shallow representations to enhance subtle cues of tiny pedestrians and suppress high-frequency noise. To establish fine-grained neighborhood correspondences under slight spatial offsets, thermal-guided local asymmetric cross-attention is designed in CCM. Finally, region-level quality and modality discrepancy are jointly modeled for adaptive cross-modal fusion in PGF. Extensive experiments on multiple UAV-based RGBT detection benchmarks demonstrate that QA2FDet achieves state-of-the-art performance and exhibits strong robustness in challenging aerial scenes.

Sensors doi: 10.3390/s26113442

Authors: Zhigang Su Jingyuan Xu Jingtang Hao Bing Han

To address the sharp degradation in positioning accuracy and the lack of robustness of received signal strength indication (RSSI)-based indoor localization methods when both the reference RSSI and path-loss exponent are mismatched, a Differential Iterative Joint Estimation (DIJE) localization method is proposed in this paper. The proposed method first employs a differential model to eliminate the uncertainty caused by reference RSSI, transforming the maximum likelihood estimation (MLE) problem into a matrix eigenvalue problem to enable fast and high-accuracy target position estimation. Additionally, an alternating iterative optimization framework for target position and path-loss exponent is constructed to achieve adaptive joint estimation of model parameters and target coordinates, effectively suppressing localization performance degradation induced by parameter mismatch. In this paper, the Cramér–Rao Lower Bound (CRLB) under the dual-parameter uncertainty scenario is derived as a theoretical performance benchmark, and both simulation experiments and public real-world datasets are used to validate the method’s performance. The results demonstrate that the DIJE method can approach the theoretical limit under varying noise levels, access point (AP) densities, and complex indoor environments. Compared with classical algorithms such as RSDPE, MLE-TLLS, SOCP3, and LCJE, the DIJE method exhibits significant advantages in localization accuracy, robustness, and adaptability to initial parameters, and can meet the engineering requirements of high-accuracy and low-latency real-time indoor localization.

Sensors doi: 10.3390/s26113441

Authors: Jiayou Wang Chaoqun Zhang Haocheng Xu Peng He

Background: Fatigued driving is a key contributing factor to major traffic accidents. Existing detection technologies suffer from issues such as delayed identification, high error rates, and a lack of quantified causal relationships between physiological and behavioral indicators. This study aims to clarify the intrinsic relationship between electrophysiological and driving behavior data during the progression of driving fatigue. Methods: Four categories of driving behavior data and electrocardiographic (ECG) heart rate variability (HRV) indicators were selected as the study subjects. Based on a four-stage standardized simulated driving experiment ranging from wakefulness to severe fatigue, the correlations between indicators were quantified using Pearson correlation analysis, and a four-layer physiological–behavioral fusion fatigue assessment model was constructed. Results: Autonomic dysregulation is the intrinsic cause of abnormal driving behavior. The two exhibit a highly synchronized, stepwise progressive evolution pattern, with |r| ≥ 0.75 among core indicators. The accuracy of the constructed model exceeded 90% for all fatigue stages, reaching 97.8% for severe fatigue detection, with a response time of ≤0.5 s. Conclusions: This model effectively addresses the limitations of single-monitoring technologies and provides theoretical support and technical guidance for multimodal identification and graded early warning of driving fatigue.

Sensors doi: 10.3390/s26113440

Authors: Haoyu Wang Yuhan Wu Xiaoming Liu Wen Li

Hand-drawn circuit diagrams remain common in education, maintenance, and early-stage design and are often photographed for storage, sharing, and reuse. Recovering electrically meaningful structure from such camera-acquired images is difficult because irregular strokes, wire discontinuities, crossings, symbol–text interference, and imaging artifacts can disrupt valid circuit topology. We therefore formulate the task as topology recovery with semantic completion rather than symbol recognition alone. To solve it, we propose a topology-consistent structural parsing framework that integrates multi-source visual perception, wire connected-component-guided connectivity reasoning, and explicit endpoint semantic recovery for direction-sensitive and multi-terminal components. On an independent benchmark of 1317 hand-drawn circuit diagrams, the proposed method achieves a 95.14% strict image-level end-to-end success rate. The recovered structures are further exported as Simulation Program with Integrated Circuit Emphasis (SPICE)-compatible netlists for downstream simulation and verification. These results support a practical vision-based image acquisition and processing workflow for converting camera-acquired hand-drawn circuit images into machine-readable and simulation-ready circuit representations.

Sensors doi: 10.3390/s26113438

Authors: A. T. M. Anishur Rahman

Electromagnetic waves are widely used including in defense, biomedicine, and fundamental science. Their efficient detection determines how we communicate, defend against adversaries, diagnose diseases and perform search and rescue operations. In this article, exploiting the precession of a levitated magnetic particle in vacuum, we show that weak electromagnetic waves down to the femtotesla level can be detected. It is also shown that such a sensor has a large dynamic range over a millitesla, is continuously tunable over many gigahertz and can detect frequencies with sub-hertz resolutions. The direction of arrival of the incoming electromagnetic wave can also be found relatively easily.

Sensors doi: 10.3390/s26113439

Authors: Tianye Luo Jing Hu Bangcheng Zhang Xinming Zhang Shaoming Luo

Reinforcement learning (RL) represents an effective approach for developing autonomous navigation and obstacle avoidance capabilities in hospital automated guided vehicles (AGVs). However, real-world adoption is challenged by the need for carefully designed reward functions, low sample efficiency, and slow convergence behaviour. To effectively address these issues, in this work, BEAGM-PPO, a reinforcement learning framework tailored for unknown hospital environments, was proposed. A reference model was initially employed to improve sample efficiency by directing the agent’s learning process. The reference model consists of expert demonstrations and policy derivation mechanisms. During the expert demonstration phase, human experts perform the required tasks and generate state-action pair datasets for training. During the policy derivation phase, demonstration data, behaviour cloning, and uncertainty estimation were used to derive the imitated expert policy. An ant colony optimization (ACO)-inspired pheromone mechanism and a memory replay strategy were incorporated to improve target-oriented action selection and supress unnecessary exploration. Experiments conducted in typical 3D simulation scenarios demonstrated that the proposed method achieved the highest arrival rate compared with baseline models. Moreover, the integrated imitation learning approach enables uncertainty estimation for both the policy and the model, while expanded training datasets further enhance performance. Overall, the results prove that BEAGM-PPO serves as a solid theoretical foundation for autonomous navigation in hospital AGVs.

Sensors doi: 10.3390/s26113437

Authors: Eray Tabak Sadullah Öztürk Arif Kösemen Necmettin Kılınç Zafer Ziya Öztürk

In this study, metal (copper, nickel, cobalt, chromium)-decorated ZnO nanorods are successfully grown on glass substrates via a hydrothermal synthesis method to test their electrical and gas-sensing properties. SEM images revealed the formation of metal nanoparticles surrounding the ZnO nanorods. To confirm that these structures originated from the metal nanoparticles, EDX analysis was performed, and the presence of metal nanoparticles was validated. XRD analysis indicated that the crystal structure of the ZnO nanorods was hexagonal, and shifts in the (002) plane were observed due to metal nanoparticle doping. ZnO nanorods functionalized with metal nanoparticles were tested at 200 °C against various gases (hydrogen, ethanol, chloroform) and at different gas concentrations. The time-dependent variation in current was observed when ZnO nanorods functionalized with metal elements were exposed to hydrogen gas at test concentrations ranging from 1000 ppm to 5000 ppm at 200 °C. The results demonstrated a clear correlation between the rate of current change and hydrogen concentration, with higher concentrations resulting in faster responses. Additionally, the sensitivity of ZnO nanorods with decorated metal nanoparticles to ethanol and chloroform gases at concentrations ranging from 1000 ppm to 5000 ppm, as well as their sensor responses to different gases at 200 °C, were also measured.

Sensors doi: 10.3390/s26113436

Authors: Ruiheng Hu Yiwei Chen Kejia Xu Leyan Zhang Chengyang Yue Hao Pi Xuhua Chen Xiaoyong Lin

UAV-based wildlife monitoring encounters tremendous challenges posed by complex environments, such as the extremely low proportion of effective targets in aerial images and variations in remote sensing scales. This paper presents a novel fusion framework named YOLIP, which integrates a detection head with semantic perception capabilities and an implicit feature adjustment module to boost detection accuracy and feature representation ability. Specifically, this paper redesigns the detection head to enable it to simultaneously learn spatial positioning and semantic features, thereby achieving more reliable extraction of regional features. The implicit feature modulation module introduces a dual-path fusion mechanism, which elevates the feature quality through geometric–semantic fusion, thereby improving the consistency and robustness of the detection. Furthermore, this paper also develops an asynchronous scheduling strategy, which can selectively execute computationally intensive operations to achieve computational optimization, enabling this framework to adapt to actual detection scenarios based on unmanned aerial vehicles. In this study, we conducted numerous experiments on the self-built drone wildlife dataset as well as the publicly available aerial wildlife dataset. Theresults demonstrate that compared with existing detection models, YOLIP improves mAP@0.5 by 11.6% while maintaining an efficient inference speed, achieving an improvement in detection performance. In addition, cross-dataset evaluation verifies the stable performance and generalization capability of the proposed method across multiple real-world scenarios.

Sensors doi: 10.3390/s26113435

Authors: Xiaojing Liao Yongwei Chi Yu Bai Qinya Dai Peiyu Zhao Na Li Linlin Sun Dongyang Li

To address the problem of fault-related structures and noise disturbances in rolling bearing vibration signals being highly coupled in the original one-dimensional signal domain under severe noise conditions, in this study, we propose a Phase-space adaptive Expert Denoising Network (PaEDNet), a robust fault diagnosis framework that integrates representation construction, adaptive restoration, and condition discrimination. Unlike existing methods that mainly enhance network modelling directly in the original signal domain, the proposed framework first constructs a spatially organised two-dimensional similarity representation through phase-space reconstruction, which further unfolds fault-related dynamic structures from temporal entanglement and provides a more suitable preliminary representation domain for subsequent restoration. On this basis, a CoPaMoE-augmented adaptive denoising module is introduced into the representation domain to improve structural restoration capability under heterogeneous noise and different local patterns. DenseNet is then employed for fault classification, thereby forming an integrated fault diagnosis framework combining representation reconstruction, noise restoration, and condition discrimination. The resulting pipeline performs end-to-end diagnosis from raw vibration signals to fault labels at inference, while training is conducted in a stage-wise manner. Experimental results derived using the two public datasets, CWRU and PU, show that the proposed method consistently outperforms multiple comparative models under different signal-to-noise ratio conditions and maintains stronger robustness in low-SNR scenarios. Under the −6 dB condition, PaEDNet achieves classification accuracies of 93.98% and 90.45% on the two datasets, respectively. Further ablation studies and expert-routing analysis demonstrate that the combination of structured representation construction and adaptive expert restoration jointly enables the improved performance of the model. In this study, we provide a new modelling perspective for the fault diagnosis of vibration signals in complex noisy environments.

Sensors doi: 10.3390/s26113434

Authors: Sezin Sayin Kristen L. Steffens Kurt D. Benkstein Mona Zaghloul Steve Semancik

Localized surface plasmon resonance (LSPR) is an optical phenomenon that occurs when light interacts with free electrons on the surface of metallic thin films, producing intensified electromagnetic fields at specific sites, often called “hot spots”. LSPR-based sensing technologies respond to chemical and associated optical interfacial changes. Inherent advantages include enhanced sensitivity, compact size, low production cost, and strong potential for integration into portable, point-of-care diagnostic systems. This study focuses on a detailed investigation into the surface functionalization of localized surface plasmon resonance (LSPR)-based nanohole array (NHA) sensors for biomedical applications. Gold-coated NHA surfaces were functionalized using polyethylene glycol (PEG) self-assembled monolayers (SAMs), enabling specific attachment of biomolecular species. As a proof-of-concept, bovine serum albumin (BSA) and SARS-CoV-2 nanobody proteins were successfully immobilized on the PEGylated surfaces. Individual steps of surface modification including PEGylation, protein immobilization and nanobody immobilization were validated through a dual-method approach which combined measurement of LSPR optical spectral shifts and x-ray photoelectron spectroscopy (XPS) chemical analyses. Reproducibility was assessed across multiple sensors and repeated trials, confirming the repeatability of each functionalization and binding process. The sensor system, consisting of NHA-based plasmonic platform, microfluidics, and a portable optical spectrometer, exhibits the capability for reliable and sensitive, label-free detection of biomolecular targets, including viral antigens, in liquid-phase environments.

Sensors doi: 10.3390/s26113433

Authors: Yoosef B. Abushark Shabbir Hassan Asif Irshad Khan

In the original publication [...]

Sensors doi: 10.3390/s26113432

Authors: Bianca Brişan Camil Jichici Raul Robu Bogdan Groza

Intrusion detection on in-vehicle networks requires high accuracy, which is reported by many papers so far, but also computational efficiency to make it suitable for real-world scenarios. The achievement of both requirements at the same time becomes harder to achieve, especially as the number of attacks diversifies. An approach to leverage computational costs is the use of sliding windows, i.e., batch processing, which extends the detection over multiple frames, but the use of multitask learning is also advantageous because a number of layers are shared between classes to extract common relevant features. While indeed the greatest computational gains are from the use of a sliding window, multitask learning has benefits too and is in fact necessary as multiple attack types can coexist in the same window. We explore the benefits of this approach on three existing attack datasets and we also build our own dataset that garners more attack complexity so that we can concretely measure the benefits of multitask learning both in terms of detection rate and computational savings. Our approach considers the feature-level similarity between attack types and legitimate frames, extracted from the mutual information between the two, and extends detection over windows of multiple frames, which justify multitask learning as frames belonging to different classes can co-exist in the same window.

Sensors doi: 10.3390/s26113431

Authors: Anak Agung Surya Pradhana Nobuo Funabiki I Nyoman Darma Kotama Kadek Suarjuna Batubulan Putu Sugiartawan

Nowadays, closed-circuit television (CCTV)cameras are deployed worldwide to monitor movements of humans and other objects to improve the efficiency and safety of societies. Therefore, their proper placement is crucial for achieving effective surveillance coverage. Additionally, their proper placement is significantly important for maximizing visual coverage while reducing installation/management costs. For this task, digital twin is a useful technology, since it can simulate coverage and blind spots while freely changing camera locations. To implement digital twin, 3D modelling of a structure including a complex room is a key issue. In this paper, we propose a 3D indoor modelling method using 360° panoramic images and show its application to a CCTV camera placement optimization. This method constructs a structured 3D model of a target room from captured 360° panoramic images using a 3D Gaussian Splatting reconstruction method based on a visual simultaneous localization and mapping (VSLAM) framework. The Inertial Measurement Unit (IMU) is used together to improve the camera position estimation accuracy. The model construction is anchored using a GNSS/GPS reference to establish global spatial coordinates. As an application of the generated 3D model, optimal locations of a given number of CCTV cameras are determined by combining ray-casting visibility analysis and a greedy optimization algorithm in the virtual environment, maximizing visual coverage while minimizing blind spots and avoiding excessive overlap between camera views. For evaluations, we applied the proposed method to three rooms in Okayama University, Japan, and seven rooms in the Indonesian Institute of Business and Technology, Indonesia. After optimizing camera locations in the virtual environment, the cameras were actually installed in the rooms according to the recommended positions. The performance was evaluated using visibility coverage, blind spot reduction, and Root Mean Squared Error (RMSE) between the estimated and actual camera positions, where promising results were achieved.

Sensors doi: 10.3390/s26113429

Authors: Xiaochang Lv Yanhui Li

Accurate characterization of reservoir core porosity and morphology from core CT images is essential for oil and gas exploration, and super-resolution offers a cost-effective means to enhance image clarity. To overcome the limitation of existing pixel-level algorithms in distinguishing high- from low-frequency information, we propose the Multi-scale Fusion Attention Mechanisms of Generative Adversarial Networks (MFAGAN). MFAGAN integrates a residual-in-residual fusion attention module to improve throat structure discrimination, a multi-scale discriminator to stabilize training and boost generator performance, and a perceptual loss to sharpen edges and textures. Comparative experiments demonstrate that MFAGAN achieves superior performance over GAN-based baselines in terms of learned perceptual image patch similarity (LPIPS), structural similarity index (SSIM), subjective image quality, and generalization ability. However, its peak signal-to-noise ratio (PSNR) is slightly lower than that of methods such as EDSR. When tested within the distribution of the DeepRock-SR dataset, MFAGAN demonstrated stable and consistent objective and subjective performance. However, its out-of-distribution generalization ability has not yet been evaluated, which is a clear limitation of the current study. In our future work, we will build upon MFAGAN to further explore 3D super-resolution reconstruction techniques for core CT images, with the aim of advancing the practical application of super-resolution methods in oil and gas exploration and development.

Sensors doi: 10.3390/s26113430

Authors: Mahyar Jafar Kazemi Maria Rashidi Won-Hee Kang Mohammad Siahkouhi

In the original publication [...]

Sensors doi: 10.3390/s26113428

Authors: Yazhou Zhou Shanshan Peng Yun Wang Nan Zhou Fei Shan

To address the issue of 2D laser-guided automated guided vehicles (AGVs) in industrial intelligent material handling scenarios being susceptible to interference from changes in lighting and complex obstacles, leading to abnormal positioning and mapping and frequent false stops, this paper designs a lightweight, multi-dimensional perception and anti-false-stop YOLOv8 anomaly recognition network, achieving accurate identification of various interferences in complex environments. An adaptive decision-making fault-tolerant control algorithm is proposed, introducing a temporal logic verification and dynamic threshold adjustment mechanism to achieve real-time dynamic switching of obstacle avoidance levels, ensuring efficient coordination between perception decision-making and control execution. An AGV anomaly detection sample set suitable for complex industrial scenarios is constructed, providing reliable data support for model optimization and accuracy evaluation. Finally, real-world deployment verification in a real electronics factory environment shows that this method reduces the vehicle false-stop rate and improves task handling efficiency. This research effectively solves the robust perception problem of AGVs in complex industrial environments and has significant engineering application value.

Sensors doi: 10.3390/s26113427

Authors: Wenbo Zhao Donghua Lu Yingjun Zhao Keyue Chen

To address misclassification and reduced accuracy in semantic change detection caused by class imbalance and variable object sizes, this paper improves BGSNet and proposes a new change detection network, SRDFNet (Semantic Refinement and Differential Features). Based on BGSNet’s framework, it introduces three complementary modules: (1) a hierarchical graph module (HGM) that converts multi-scale feature maps into compact semantic graph nodes, using graph attention for intra-layer and cross-level semantic interaction to enhance topological relationship perception; the HGM mitigates the effects of class imbalance by compacting multi-scale features into semantic nodes; (2) a difference enhancement (DE) module that extracts multi-receptive-field difference information from bi-temporal concatenated features via multi-scale parallel convolution branches; (3) a semantic refine (SR) module that performs lightweight residual refinement on bi-temporal semantic features to improve the segmentation accuracy. The DE and SR modules mitigate the degradation in semantic segmentation accuracy caused by variable object sizes. It is trained and tested with BGSNet and three other models on the SECOND and HRSCD datasets. For the SECOND dataset, in terms of five quantitative indicators, namely OA, mIoU, SeK, F1 and recall, SRDFNet achieves 87.64%, 70.31%, 20.36%, 60.25% and 65.27%, respectively. Compared with BGSNet, it gains performance increases of 1.34%, 0.73%, 1.44%, 0.81% and 2.72%, respectively. For the HRSCD dataset, SRDFNet achieves 98.13% (OA), 52.67% (mIoU), 73.77% (SeK), 88.86% (F1) and 88.18% (recall), ranking first among the four methods. Compared with BGSNet, it gains performance increases of 3.96%, 3.93%, 9.69%, 2.33% and 4.00%, respectively.

Sensors doi: 10.3390/s26113426

Authors: Lin Xiao Chengli Zhao

During the hoisting of prefabricated segmented beams, the alignment of rods and holes mainly relies on manual operation, which suffers from low safety and efficiency. To improve the safety and efficiency of rod–hole alignment, this paper proposes a vision-based alignment method for hoisting prefabricated segmented beams. The method uses binocular vision to measure the spatial coordinates of key points on rods and holes, establishes a mathematical model for alignment, and calculates the center distance and relative rotation angle between them. An experimental platform is built and tests are conducted. The results show that the proposed method can effectively measure the center distance and rotation angle, improve measurement efficiency and safety, achieve high accuracy, and possess high practical engineering value.

Sensors doi: 10.3390/s26113425

Authors: Eytan Kats Christoph Grossbroehmer Ziad Al-Haj Hemidi Fenja Falta Wiebke Heyer Mattias P. Heinrich

Medical image registration is a fundamental task in medical image analysis, enabling the alignment of images from different modalities or time points. However, intensity inconsistencies and nonlinear tissue deformations pose significant challenges to the robustness of registration methods. Recent approaches leveraging self-supervised representation learning show promise by pre-training feature extractors to generate robust anatomical embeddings, that further used for the registration. In this work, we propose a novel framework that integrates equivariant contrastive learning directly into the registration model. Our approach leverages the power of contrastive learning to learn robust feature representations that are invariant to tissue deformations. By jointly optimizing the contrastive and registration objectives, we ensure that the learned representations are not only informative but also suitable for the registration task. We evaluate our method on abdominal and thoracic image registration tasks, including both intra-patient and inter-patient scenarios. Experimental results demonstrate that the integration of contrastive learning directly into the registration framework significantly improves performance, surpassing strong baseline methods.

Sensors doi: 10.3390/s26113420

Authors: Lu Zhang Yi Du Haolong Li Shiquan Yan Quanxing Yao Chunyu Liu Yuejun Zhang Xiaojian Zhu

Driven by the profound convergence of the Internet of Things (IoT) and ubiquitous computing, wearable multifunctional sensors have emerged as a key technology for high-precision human activity recognition (HAR). Advancements in novel materials and flexible electronics have propelled the evolution of these sensors, enabling advances in decoupling heterogeneous signals, enhancing system robustness, and expanding environmental perception. This review systematically examines the frontier research on wearable multifunctional sensors for HAR. We provide an in-depth analysis of three core architectural design paradigms: architecture-level integration, which relies on physical spatial isolation for hardware-level signal decoupling; monolithic integration, which strives for extreme spatial compactness and spatiotemporal signal consistency; and the emerging intrinsically multifunctional design, which leverages novel stimuli-responsive materials for the intrinsic orthogonal discrimination of multidimensional signals. Furthermore, we delineate the diverse application scenarios of these highly integrated sensing platforms across medical rehabilitation, sports science, human–computer interaction (HCI), and daily behavior perception. Finally, this article discusses the critical challenges currently confronting this technology and outlines its future development prospects.

Sensors doi: 10.3390/s26113424

Authors: Rushang Zhang Xiaogang Fu

The increasing use of UAVs has raised concerns regarding public safety and airspace security. To address air-to-air micro-UAV detection with cluttered backgrounds, tiny targets, and diverse viewing angles, this paper develops SPAE-YOLOv8, a lightweight detector based on YOLOv8n. SPAE consists of four core designs: SIoU loss, P2 shallow feature layer, ADown adaptive downsampling, and Efficient_UAVDet lightweight detection head. These modules improve small-target representation and reduce model size. In this paper, lightweight refers to the combination of parameter count, storage volume and inference speed. On the Det-Fly dataset, the proposed method achieves an mAP@0.5 of 0.922, outperforming YOLOv8n by 7.2 percentage points while reducing total parameters by 30%. We conduct independent training and testing on the DUT Anti-UAV dataset and obtain an mAP@0.5 of 0.906. Cross-dataset testing is further carried out on the more challenging Anti-UAV300 dataset without additional fine-tuning to verify the generalization performance of the model. In real-world onboard deployment, the model is implemented on an Intel NUC11TNHi7 embedded UAV platform with OpenVINO acceleration and achieves 43.9 FPS at a resolution of 640×640, satisfying real-time inference requirements. The ablation results demonstrate the contribution of the proposed modules, providing an efficient lightweight solution for airborne monitoring and civil airspace security.

Sensors doi: 10.3390/s26113423

Authors: Yifeng Huang Xikang Lu Daode Zhang

Tool wear state monitoring under varying operating conditions is important for machining quality and production reliability. However, changes in cutting parameters can shift monitoring-signal distributions and reduce the generalization ability of data-driven models. This paper proposes a cross-condition tool wear state monitoring method based on multi-source sensor signal fusion and supervised transfer learning. X-axis vibration, Z-axis vibration, and spindle current signals are organized as multi-channel time-series inputs. A deep model integrating a multi-scale convolutional neural network, bidirectional long short-term memory, and an attention mechanism is developed to extract discriminative wear-related features. Source-domain pretraining, target-domain warm-up fine-tuning, and source-target joint fine-tuning are organized as a progressive supervised transfer procedure to improve target-condition adaptation. Experiments are conducted on a custom multi-condition dataset using an hp0 + hp1 → hp2 transfer task. Under the unified XZI input configuration, the proposed method outperforms CNN-LSTM, DANN, and CORAL. Input ablation results show that X, XZ, and XZI achieve accuracies of 0.6000, 0.7647, and 0.8588, respectively. In repeated random-seed experiments, the method obtains an Accuracy of 0.7929 ± 0.0499, a Macro-F1 of 0.7292 ± 0.0706, and a Cohen’s Kappa of 0.6542 ± 0.0840. The results demonstrate the effectiveness of multi-source sensor fusion and supervised target-condition adaptation for cross-condition tool wear monitoring.

Sensors doi: 10.3390/s26113422

Authors: Yuyu Sun Zongkai Shang Mingxiao Yang Fandi Meng Mengxuan Mu Heqi Yan

Accurate LiDAR point cloud registration on resource-constrained edge platforms is a prerequisite for intelligent robotics and industrial automation, yet it remains challenging because low-overlap matching, false correspondences, and fine alignment must be handled under limited computing budgets without GPU acceleration. While learning-based methods have advanced the field, their heavy hardware dependency and training requirements often hinder their practical deployment on mobile edge devices. To bridge this gap, this paper proposes GeoRescue, a training-free geometric registration framework designed for high-precision perception under stringent hardware limits. The method consists of three modular stages: Asymmetric Correspondence Expansion (ACE), which enlarges the candidate correspondence set to reduce the loss of true matches; Dynamic Geometric Topology Gating (DGTG), which suppresses false matches through distance-consistency-based hypothesis filtering; and Uncertainty-Aware Manifold Refinement (UAMR), which improves fine alignment by explicitly modeling local anisotropic noise via covariance-guided optimization. Experiments on 3DMatch, 3DLoMatch, and KITTI show that GeoRescue achieves registration recall rates of 84.84% and 41.27%, respectively, and a 94.95% success rate on KITTI. Remarkably, the framework matches the accuracy of high-capacity learning models while running on a GPU-free, 15 W edge CPU platform (Intel Core i5-8265U). These results indicate that GeoRescue provides a deployment-ready solution with an optimal efficiency–accuracy trade-off for LiDAR sensing and robotics perception in complex, real-world scenarios.

Sensors doi: 10.3390/s26113421

Authors: Seth Donahue J.D. Peiffer R. Tyler Richardson Yishan Zhong Shaun Q. Y. Tan Benoit L. Marteau Stephanie A. Russo May D. Wang R. James Cotton Ross Chafetz

This study validates a clinically accessible approach for quantifying the Upper Extremity Reachable Workspace (UERW) using monocular AI-driven Markerless Motion Capture (MMC). Objective validation of such techniques for clinically oriented tasks is essential to support their adoption in clinical motion analysis. Nine adults without impairments performed the standardized UERW task, reaching targets distributed across a virtual sphere centered on the torso and displayed via VR headset. Movements were simultaneously captured with a marker-based system and eight FLIR cameras; monocular analysis was applied to two videos representing frontal and offset camera configurations. Agreement was assessed by comparing the percentage workspacereached across six of eight workspace octants between the systems. The frontal camera demonstrated strong agreement with the marker-based reference (mean bias: 0.61±0.12% reachspace per octant), whereas the offset view underestimated workspace reached −5.66±0.45%. Depth-related errors in the frontal configuration were confined to posterior octants, whereas the offset view introduced inaccuracies in both contralateral and posterior octants. These findings support the feasibility of a frontal monocular camera for UERW assessment, particularly for anterior workspace evaluation. While posterior accuracy remains limited by depth estimation and anatomical occlusion errors, the overall results demonstrate clinical potential for practical, monocular-camera assessments.

Sensors doi: 10.3390/s26113419

Authors: Zhiqi Li Yajun Chen Mingrui Li Ruipeng Wang Hao Liu

The rapid evolution of deepfake generation techniques has made high-fidelity facial manipulation a critical threat to social credibility and personal privacy, demanding detection algorithms with strong cross-domain generalization. Existing methods suffer from two fundamental limitations: spatial-domain approaches cannot capture imperceptible forgery artifacts, while frequency-aware methods lack effective integration of spatial semantic and spectral features. To address these challenges, we propose FreqMamba, an end-to-end face forgery detection framework that adaptively aggregates spatial semantic features and frequency-domain artifacts via a gated late-fusion mechanism, and performs global sequence modeling using a bidirectional vision state space model (Vim). FreqMamba consists of three core components: a CNN branch for spatial semantic features, a hierarchical discrete wavelet transform (DWT) branch for fine-grained frequency artifacts, and a bidirectional Mamba backbone for global sequence modeling with linear complexity. The gated fusion mechanism adaptively combines multi-branch features, enhancing responses in forgery-rich regions while suppressing irrelevant noise. Trained exclusively on FaceForensics++ (c23), FreqMamba achieves strong cross-domain performance: on Celeb-DF v2, it attains 0.7767 AUC, surpassing a comparable-parameter CNN baseline (1.14 M parameters, 0.7262 AUC) by 5.05 percentage points; on the real-world WildDeepfake dataset, it achieves 0.6993 AUC, outperforming the lightweight CNN baseline (0.6272 AUC) by 7.21 points. Ablation studies confirm that DWT frequency priors and Mamba sequence modeling exhibit synergistic effects, and Grad-CAM visualizations validate the model’s focus on critical forgery regions. FreqMamba provides an effective approach for generalized face forgery detection in cross-domain scenarios.

Sensors doi: 10.3390/s26113418

Authors: Shiyuan Zheng Zhaochao Li

Metal pipelines may have various defects due to long-term service, corrosion, external strikes, etc. Traditional closed-circuit television (CCTV) inspection techniques are capable of detecting these defects. However, substantial human resources are required and the detection results are subjected to human subjectivity. Thus, this study develops a deep learning-based intelligent defect detection model for metal pipeline images. CNNs (convolutional neural networks) are utilized to automatically extract defects, which may mitigate the interference of subjective factors and enhance the recognition capability of the defects in pipelines. The proposed model builds upon the original YOLOv8n model by incorporating the SCSA (Spatial and Channel Synergistic Attention) mechanism, LskBlock, and SlideLoss function, respectively. These enhancements improve the ability to detect small targets, increase recognition accuracy, and facilitate global optimization, respectively. The developed YOLO-LSS (YOLOv8n-LskBlock-SlideLoss-SCSA) model is compared with other deep learning models characterized by the following metrics: mAP50, mAP50:95, precision, recall rate, and F1-score, respectively. It is found that mAP50 achieves 79.05% (+2.86%), mAP50:95 53.7% (+7.19%), precision 81.6% (+5.30%), recall rate 75.5% (+2.90%), and F1-score 78.4 (+4.01), indicating that the proposed model effectively enhances the capability of detecting internal defects in pipelines.

Sensors doi: 10.3390/s26113415

Authors: Su-Ha Lee Chang Ho Song

Older adults with mild cognitive impairment (MCI) often show gait impairment during dual-task walking and obstacle negotiation. This assessor-blinded randomized controlled trial investigated whether dual-task gait training with visual adaptation, added to a general exercise program, improves gait and related functional outcomes in older adults with MCI. Forty participants aged 65 years or older who met the MCI criteria were randomly allocated to a dual-task gait training with visual adaptation group or a control group (n = 20 each). Spatiotemporal and adaptive gait parameters were assessed before and after 4 weeks of intervention during level walking and during predictable and unpredictable obstacle negotiation under light and noise conditions. Balance, executive function, and concern about falling were also evaluated. Compared with the control group, the intervention group showed greater improvements in level walking and predictable obstacle negotiation, including longer step and stride length, shorter step and stride time, higher cadence, and faster gait speed. Under unpredictable obstacle conditions, gains were more selective and were observed mainly in step and stride length and adaptive gait indices. The intervention group also showed greater improvement in balance and executive function and a larger reduction in concern about falling. These findings suggest that adding dual-task gait training with visual adaptation to a general exercise program may have clinical value for improving adaptive gait and related functional outcomes in older adults with MCI. However, because the intervention group received additional gait-specific training and a higher total training dose than the control group, future dose-matched studies are needed to clarify the specific contribution of visual adaptation.

Sensors doi: 10.3390/s26113417

Authors: Jahidul Islam Dristi Datta Fowzia Akhter

Heart disease remains one of the leading causes of mortality worldwide, creating a strong need for reliable population-scale risk prediction models for large-scale screening and preventive monitoring. However, existing machine learning and deep learning approaches often struggle under severe class imbalance, data leakage risks, and unstable precision–recall trade-offs, limiting reliability in population-scale health-monitoring settings. To address these challenges, this study proposes a precision-aware Dual-Branch FT-Transformer framework for cardiovascular risk prediction using the BRFSS-2024 dataset. The proposed architecture separates recall-oriented detection and precision-oriented verification through two specialized prediction heads and integrates them using a lightweight gating mechanism trained strictly within training folds to prevent information leakage and enable controlled error arbitration. Under a strict leakage-safe 5-fold cross-validation protocol, the proposed model achieves an F1-score of 0.43, recall of 0.59, and AUPRC of 0.38 at a fixed threshold of 0.50 while reducing false negatives by more than 50% compared to LightGBM without excessive false positives. Although some baseline models achieve higher AUROC values, the proposed framework demonstrates more balanced and clinically meaningful precision–recall behaviour at operational screening thresholds. Additional evaluation on an independent NHANES cohort under the same leakage-safe re-training protocol further suggests robustness across heterogeneous population-health settings. Overall, the proposed dual-objective learning framework provides a practical and robust approach for imbalanced tabular prediction in population-scale cardiovascular risk assessment.

Sensors doi: 10.3390/s26113416

Authors: Federico Orozco-Santos Víctor Sempere-Payá Javier Silvestre-Blanes

Wireless networks are increasingly considered for industrial and time-critical applications, where flexible deployment must be reconciled with predictable communication behaviour. IEEE 802.11ax introduces mechanisms such as Orthogonal Frequency Division Multiple Access (OFDMA), Trigger-based Uplink Access (TUA), and Target Wake Time (TWT) as part of ongoing efforts to support bounded latency and deterministic transmissions in Wi-Fi networks. However, the practical behaviour of these mechanisms depends not only on the standard, but also on what commercial devices expose, how access points implement scheduling decisions, and how trigger-based access, RU assignment, and timing control can be configured in real deployments. This paper therefore focuses on the practical implementation and experimental assessment of OFDMA-based deterministic operation using Wi-Fi 6 commercial off-the-shelf (COTS) hardware. The proposed configuration combines driver-level enabling of high-efficiency mechanisms with controlled testbed measurements and complementary simulations, allowing OFDMA operation to be compared against conventional single-user OFDM under realistic traffic and interference conditions. The results show that coordinated OFDMA operation on COTS devices improves temporal stability, reducing jitter by up to 23% and latency by approximately 44% with respect to single-user OFDM operation. The experiments also reveal practical effects that are central to deterministic-oriented Wi-Fi: simultaneous RU-based transmissions reduce contention-driven variability, TWT-based activity windows improve temporal alignment, and RU subdivision introduces a throughput trade-off that must be considered when dimensioning industrial traffic. Overall, the study provides empirical evidence that Wi-Fi 6 can support deterministic-oriented industrial communication when OFDMA, trigger-based access, and timing mechanisms are jointly configured, while also highlighting the implementation constraints that remain when moving from standard capabilities to COTS device behaviour.

Sensors doi: 10.3390/s26113414

Authors: Hongbo Yu Yu Liu Jiadi Wei

In remote areas where sugar beets are grown on a large scale, inadequate ground-based communication networks can easily lead to information silos in farmland, as well as technical challenges such as uneven node power consumption and short lifespans during the long-term operation of wireless sensor networks. To address these challenges, a real-time field environment monitoring system for sugar beet fields based on the Beidou satellite system and ZigBee wireless sensor networks has been developed, employing a three-tier architecture comprising a perception layer, a network layer, and an application layer. The system uses ARM as the core of the data acquisition nodes and integrates sensors for temperature, humidity, light intensity, atmospheric pressure, and dissolved oxygen with a Beidou positioning module. Field data are aggregated via a ZigBee mesh network and transmitted remotely using a dual-link Beidou short message protocol. To prevent uneven energy consumption in ZigBee networks, an improved energy-balanced routing algorithm, Energy-Balanced Low-Energy Adaptive Clustering Hierarchy (EB-LEACH), is proposed. By optimizing cluster head election, adaptive competition radius mechanisms, and inter-cluster multi-hop routing strategies through multi-factor weighting, the algorithm achieves a globally balanced distribution of network energy consumption. Our experimental tests demonstrate that, compared to the traditional LEACH protocol, this algorithm increases the number of rounds until the first node fails by 87.3%, extends the network half-life by 110.48%, and improves total packet delivery by 118.3%. Our test results indicate that the improved routing algorithm performs better, and the accuracy of the sensor measurements meets the practical requirements for environmental monitoring in sugar beet fields.

Sensors doi: 10.3390/s26113413

Authors: Jiahao Li Renjie Li Nan Wang

Efficient coordination under limited communication is a central challenge in multi-agent reinforcement learning (MARL). Existing approaches often focus on message exchange without explicitly modeling how communication affects policy learning, leading to redundant interactions and limited coordination gains. In this paper, we propose a threshold-gated sparse communication framework built upon QMIX, a monotonic value-decomposition method that mixes individual agent action values into a global team action value. In the proposed framework, communication is integrated into the agent utility function to directly influence policy learning. Each agent encodes local observations into structured representations and activates communication through a learned trigger mechanism. Messages are aggregated via neighbor-constrained attention and incorporated into utility estimation for decentralized decision-making. Experimental results on the StarCraft Multi-Agent Challenge (SMAC) benchmark show that the proposed method improves coordination quality and training stability while significantly reducing communication frequency. On MMM, the Marine–Marauder–Medivac heterogeneous scenario, the communication rate is reduced to approximately 30–38% while achieving up to 96.6% win rate, compared to 92.1% for QMIX. On 10m_vs_11m, a homogeneous scenario where ten allied Marines fight against eleven enemy Marines, communication remains within 28–37% while reaching 88.4% win rate, compared to 85.6% for QMIX. Moreover, on the same task, varying communication thresholds induce clearly differentiated policy behaviors, indicating that sparse communication not only reduces overhead but also plays a critical role in shaping coordination policies. These results demonstrate that selective communication enables efficient coordination while explicitly regulating policy formation.

Sensors doi: 10.3390/s26113410

Authors: Stamatios Baltzis Gerasimos Pagiatakis Nikolaos Voudoukis Andreas Papadakis Leonidas Dritsas Dimitris Uzunidis

This article describes a portable Arduino-based prototype for the recording and analysis of electroencephalogram (EEG) signals associated with anxiety situations. The system’s main aim is to enable the user to self-detect stress and take self-regulating/relaxing actions in real time before stress escalates. The recorded EEG signals are first processed in the analog domain (including amplification and noise reduction) and then, by using an Arduino Uno board, they are converted into digital format and transmitted through either a wired or wireless connection to a computer to be depicted in both the time and the frequency domains by means of an open-source software. During the performed tests, the system successfully showed visible changes in the alpha and beta brain signals corresponding to the states of resting, induced stress, and the subsequent self-regulation/relaxation process. The proposed prototype (though non-clinical in its present form) has the merits of relatively low cost, easy self-use (outside clinical environments), and real-time EEG signal depiction, and, apart from enabling the user to self-detect and self-monitor stress, it can also be used for educational and/or research purposes.

Sensors doi: 10.3390/s26113412

Authors: Yu Ji Zixuan Liu Xin Gu Chenze Huo Zihan Zhang Song Tang Jun Mei Can Huang

Photovoltaic (PV)-storage systems operating in weak grids are affected by high grid impedance, transient voltage disturbances, and measurement noise, which can degrade frequency regulation, increase converter current stress, and impose high-frequency current fluctuations on the battery. To address these issues, this paper proposes a multi-timescale transient-state sensing and signal-processing framework for grid-forming PV-hybrid storage systems. The proposed framework combines three coordinated functions. First, a frequency-domain HESS power-decoupling mechanism separates high-frequency transient power components and assigns them to the supercapacitor, while the battery mainly handles low-frequency energy variations. Second, a voltage-deviation-driven adaptive virtual inductance is introduced to increase the equivalent output impedance during voltage-sag events and reduce transient inrush current. Third, a noise-resilient frequency sensing strategy based on a filtered frequency derivative and a dead-band for false-trigger suppression is developed to reduce noise-induced false triggering in adaptive inertia and damping control. Comparative simulations indicate that under the tested weak-grid conditions, the proposed method reduces the transient inrush-current peak by 53.2%, decreases the maximum dynamic frequency deviation by approximately 75%, and improves the active-power regulation speed by more than 50%. These results indicate that the proposed sensing-oriented framework can improve transient response while reducing converter and battery current stress in PV-storage systems connected to high-impedance grids.

Sensors doi: 10.3390/s26113411

Authors: Borui Zhang Mengyao Sun Jie Chao

DNA nanotechnology exploits the high precision and programmability of base complementary pairing to construct diverse nanostructures. Especially, the integration of stimuli-responsive modules has driven the evolution of DNA nanostructures toward dynamic nanomachines, conferring significant advantages in biomedical applications. In this review, the stimulus–response strategies of DNA nanostructures are first outlined, including molecular-driven mechanisms and environmental stimulation methods. The recent advances in the application of nanomachines based on various response strategies in biosensing and drug delivery are then elaborated. Finally, current challenges and prospects for these nanomachines in clinical diagnosis and precision therapy are discussed. This review provides a systematic reference for the development of responsive DNA nanomachines and the exploration of their biomedical applications.

Sensors doi: 10.3390/s26113409

Authors: Dajie Tian Levent Guvenc

Automated valet parking requires reliable long-range slot searching and precise low-speed docking in confined structured lots. This paper proposes a hierarchical cruise-to-park framework that combines nonlinear model predictive control (NMPC) for predefined-route cruising with a Twin Delayed Deep Deterministic Policy Gradient (TD3) agent for terminal parking. The system is implemented in a structured Simulink environment with Unreal Engine-based geometry-aware sensing modules. During cruising, a camera-based module detects available slots and triggers the transition to parking. The NMPC uses a custom cost function to improve tracking on curved approaches, while the TD3 policy uses LiDAR feedback and reward shaping with an explicit time penalty to encourage efficient, stable docking. Simulation results demonstrate smooth phase transition, accurate cruising, and effective terminal parking in the training slot. Validation on six previously unseen target slots within the same parking-lot environment shows encouraging intra-lot target-slot transferability without retraining. Additional PPO and SAC comparisons and a time-penalty ablation further evaluate the relative learning performance and the effect of reward design, supporting the proposed architecture as a practical baseline for integrated cruise-to-park automated valet parking studies.

Sensors doi: 10.3390/s26113408

Authors: Enci Fan Tao Xie Fan Wu

The accuracy of transit-time ultrasonic flow measurement depends strongly on stable time-of-flight (TOF) extraction. However, threshold-based TOF methods are susceptible to emission transients, structural ringing, and repeated threshold crossings, which may cause false triggering and timing fluctuations. This paper proposes a time-of-flight extraction method based on time-sequenced pulses for ultrasonic flow measurement using an FPGA-based time-to-digital converter (TDC). The method equalizes signal input paths, combines peripheral path switching with FPGA gating to achieve windowed valid-edge extraction, and exploits the temporal correspondence among consecutive excitation pulses to construct multiple TOF observations from matched pulse pairs, thereby improving extraction stability and timing efficiency. A complete FPGA-TDC ultrasonic flow measurement platform based on a Zynq UltraScale+ device was developed, and TDC linearity, timing precision, and flow calibration experiments were conducted. After code-density calibration, the four channels achieved RMS values of about 20 ps, and within a flow range of 0.7–3.6 m3/h, the relative error with respect to the reference flowmeter remained within ±0.6%, with repeatability errors below 0.3%. The platform operated stably under the present experimental conditions. These results demonstrate improved TOF extraction stability and overall flow measurement performance without complex full-waveform processing.

Sensors doi: 10.3390/s26113407

Authors: Ben D. M. Jones Jon Wheat Kane Middleton David L. Carey Ben Heller

Stride-time variability is typically quantified using measures including the coefficient of variation (CV) and detrended fluctuation analysis (DFA). Whilst several studies have reported guidelines for applying such measures, few have considered the sensitivity of these measures to the way stride times are obtained. This study investigated the agreement and between-day reliability of stride-time variability measures derived from wearable devices compared to measures derived from an instrumented treadmill. Thirty-one runners completed eight minutes of running on two days. Stride times were obtained concurrently using Loadsol® insoles, Blue Trident inertial measurement units (IMUs) at four sampling rates, RunScribe™ IMUs and an AMTI instrumented treadmill. Stride-time CV and DFA-α were calculated on time-matched series. Agreement with the instrumented treadmill was quantified using Bland Altman plots and concordance correlation coefficients. Loadsol® insoles and RunScribe™ IMUs displayed the highest and lowest agreement, respectively. For the Blue Trident IMU, sampling below 400 Hz reduced agreement relative to higher rates. Between-day reliability was moderate-to-good and poor-to-moderate for stride-time CV and DFA-α, respectively, although within narrower bands than reported across studies using different measurement devices. Hence, Loadsol® insoles and Blue Trident IMUs, at sufficient sampling rates, can facilitate stride-time variability analyses, although changes over time should be interpreted cautiously.

Sensors doi: 10.3390/s26113406

Authors: Youhui Cui Zhang Zhang Liang Chang

Human pose estimation in orbit is critical for astronaut health monitoring, task assistance, and intelligent human–robot interaction aboard space stations. However, in microgravity, human poses exhibit arbitrary orientations and are often affected by severe occlusion and complex background interference, while the scarcity of annotated in-orbit data makes it difficult to directly transfer models trained on ground-based datasets. Existing semi-supervised methods also lack explicit constraints from human structural topology and pose-related physical priors, which often leads to unreasonable pseudo-labels and limits performance gains. To address these issues, we propose a physics-inspired semi-supervised pose estimation framework for microgravity scenarios. Specifically, a Canonical Orientation Constraint is introduced to alleviate orientation ambiguity; a Structure-aware Pseudo-Label Refinement module is designed to improve pseudo-label quality; and an Uncertainty-guided Rotational Consistency Framework is proposed to adaptively weight consistency learning under multi-view rotation augmentation. Within a Mean Teacher architecture, the proposed method jointly optimizes the supervised loss, orientation constraint, pseudo-label refinement, and rotational consistency objectives. Experiments on the Astro-Pose dataset show that the proposed method consistently outperforms both fully supervised and semi-supervised baselines under various extreme poses and occlusion conditions, improving AP from 47.6 to 55.6 and AR from 52.4 to 60.1, demonstrating its potential for space-station visual monitoring.

Sensors doi: 10.3390/s26113404

Authors: Jiacheng Shi Wenjun Luo Xuemei Wang Jian Guo Hengyi Ren

Fruit detection and maturity recognition are crucial for intelligent tomato harvesting and management. However, in complex field environments, challenges such as the similarity in color between fruits and leaves, cluttered backgrounds, and severe occlusions significantly hinder accurate tomato detection. To address these issues, this paper proposes a lightweight tomato maturity detection model, termed FDA-YOLO. Building upon the YOLOv11 framework, the proposed model enhances global perception in complex scenarios by introducing a multiscale feature enhancement module. In addition, a foreground–background dual-path attention mechanism is designed to better distinguish fruits from the background, thereby improving detection robustness. Furthermore, a lightweight asymmetric detection head is constructed to reduce computational cost while maintaining high accuracy. These improvements enable the model to achieve more efficient and accurate tomato maturity detection under complex conditions. Extensive experiments are conducted on the LaboroTomato dataset. The results demonstrate that FDA-YOLO achieves the best performance with relatively low computational overhead, reaching 83.4% and 67.5% in mAP50 and mAP50–95, respectively, while also attaining a near-optimal F1 score. Overall, the proposed model achieves an excellent balance between accuracy and efficiency, providing an effective solution for intelligent agricultural monitoring and automated harvesting systems.

Sensors doi: 10.3390/s26113405

Authors: Asma Komal Shuaiyong Li

As the Internet of Things (IoT) grows, strong, scalable, and adaptive intrusion detection systems (IDS) become increasingly critical for protecting IoT environments. This paper presents a comprehensive and systematic survey of IDS techniques for IoT environments, covering literature from 2021 to early 2026. The review introduces a multidimensional taxonomy that categorizes IDS approaches by detection strategy, learning paradigm, deployment architecture, and evaluation methodology. We examine conventional techniques, such as signature-based and anomaly-based detection, as well as modern machine-learning and deep-learning approaches. Furthermore, emerging paradigms, including Federated Learning, Explainable AI (XAI), TinyML, Large Language Models (LLMs), Transformer, Quantum Machine Learning, Generative Adversarial Networks and Incremental Learning, are analyzed with respect to their applicability to resource-constrained IoT environments. The paper also provides a detailed analysis of publicly available IDS datasets, validation protocols, and evaluation metrics used for benchmarking detection systems. In addition, critical challenges, including dataset realism, adversarial robustness, scalability, privacy preservation, and ethical considerations, are discussed. Finally, we highlight open research directions and propose guidelines for designing next-generation, trustworthy, and scalable IDS frameworks for IoT networks.

Sensors doi: 10.3390/s26113403

Authors: Zhangyi Yang Zuoheng Liu Wei Dong

We demonstrate a reconfigurable microwave frequency measurement (MFM) scheme based on the period-one (P1) dynamics of an optically injected semiconductor laser. Unlike conventional architectures relying on electrical frequency-sweeping, our approach utilizes the P1 oscillation to generate a wideband linear optical chirp. A spectral gating mechanism is introduced, where an optical bandpass filter creates a negative temporal marker by rejecting free-running component of distributed feedback laser (DFB), thereby eliminating the need for external synchronization or pilot tones. The measurement range is flexibly tunable by adjusting the injection parameters, enabling a measurement range from 10 to 48 GHz. Experimental results demonstrate a frequency resolution of 50 MHz with chirp rate of 1 GHz/μs and a root-mean-square (RMS) error below 15 MHz, confirming the validity of this all-optical, self-referenced frequency-to-time mapping technique.

Sensors doi: 10.3390/s26113402

Authors: Yan Zhang Xiaoyu Gong Xiaoyang Yuan

Hybrid brain–computer interfaces (BCIs) have attracted growing research attention because they combine the millisecond-level temporal resolution of electroencephalography (EEG) with the spatially informative hemodynamic responses of functional near-infrared spectroscopy (fNIRS). However, most existing deep fusion methods rely on static late-fusion strategies, which tend to underexploit latent cross-modal dependencies and are vulnerable to modality-specific signal degradation. To address these limitations, we propose MGFNet, a multi-granularity fusion network for hybrid BCI decoding. MGFNet contains three components: (1) intra-modal encoders that learn modality-specific spatiotemporal representations from EEG, oxygenated hemoglobin (HbO), and deoxygenated hemoglobin (HbR) signals; (2) cross-modal interaction encoders that temporally align paired modalities and use dilated convolutions to capture long-range EEG-fNIRS dependencies; and (3) a Coupling-Guided Sparse Component Routing (CGSCR) module that estimates sample-specific cross-modal coupling and performs adaptive discrete routing. We further introduce a deep supervision strategy to stabilize optimization and improve branch-level discriminability. Under a within-subject held-out evaluation protocol on a public benchmark dataset, MGFNet achieved classification accuracies of 99.40% on the n-back task and 99.03% on the word generation (WG) task, outperforming representative comparison methods evaluated under a matched protocol. Ablation studies further confirmed the contributions of the intra-modal encoders, the cross-modal interaction encoders, and the CGSCR module. Under controlled EEG corruption with additive white Gaussian noise at −10 dB, MGFNet outperformed a static-fusion variant by 9.23 percentage points on the n-back task and 6.31 percentage points on the WG task. These results support the effectiveness of MGFNet in the present offline within-subject setting and indicate improved robustness under controlled single-modality degradation.

Sensors doi: 10.3390/s26113401

Authors: Erdal Özbay Feyza Altunbey Özbay

Accurate and efficient semantic segmentation of point cloud data is critical in many application areas involving indoor scene understanding. In particular, fine-grained object categories, high data density, and class imbalance in high-resolution indoor datasets significantly limit class discrimination in 3D semantic segmentation. The multimodal data structure, high-fidelity geometry, and long-tail class distribution of the recently popular ScanNet++ dataset further exacerbate these challenges. This study proposes a novel Long-Tail Aware Cross-Modal Graph Attention Network (LT-CM-GACNet++) to address fine-grained 3D semantic segmentation under long-tail distributions. The proposed method integrates dynamic graph-based geometric feature extraction with a lightweight visual feature extractor based on MobileNetV3, enabling effective fusion of geometric and RGB-based information. The proposed Cross-Modal Graph Attention (CMGA) module facilitates adaptive information transfer between modalities, enabling more effective representation learning of both local and global contextual features. To mitigate the adverse effects of long-tail class distributions, prototype-based representation learning and a class frequency-aware loss function are jointly employed. This strategy improves the learning of rare classes while enhancing the discrimination between visually and geometrically similar categories. In the preprocessing stage, density-based sampling, normal vector estimation, and block-based fixed-size point cloud generation are applied to high-resolution mesh-derived data. The proposed model is evaluated on 50 scenes and 100 semantic classes selected from the ScanNet++ dataset. Experimental results demonstrate that the proposed method achieves significant improvements over existing approaches in terms of both overall segmentation performance and rare-class performance. In particular, notable gains are observed in mean Intersection over Union (mIoU) and rare-class mIoU metrics. These results highlight the effectiveness of cross-modal learning for high-resolution 3D scene segmentation under long-tail distributions.

Sensors doi: 10.3390/s26113400

Authors: Yikang Chen Hongyuan Chen Yuan Yin Junyi Chen Bo Lu Tao Chen Minglu Zhu

Urban pipelines are essential infrastructure components in modern cities. Their curved and confined structures make sensing difficult to achieve. In conventional flexible sensing devices, pressure and bending signals often interfere with each other. To address this problem, we propose an integration strategy for multi-array sensors on flexible printed circuits. The approach integrates laser-induced graphene pressure sensors and bending sensors on a polydimethylsiloxane substrate with flexible printed circuits. This integration enables stable and reliable signal acquisition and the device shows good performance under pressure loading. It has high linearity (R2 > 0.99), low hysteresis (2.68%), and a fast response time (~50 ms) in the range of 0–120 kPa. The sensing architecture is based on geometry-induced strain-field differentiation, which suppresses pressure–bending cross-interference and improves multimodal signal discrimination through structural design. Pressure mainly produces isotropic signals, while bending generates anisotropic strain signals. We test the device in simulated pipeline environments. Protrusion defects and corrosion defects generate different signal patterns. These differences allow clear defect identification. The device further supports spatial posture sensing and bending-state monitoring in complex curved pipeline conditions.

Sensors doi: 10.3390/s26113398

Authors: Jinze Li Zeling Wen Zhaoke Ning

Real-time cooperative control of quadrotor swarms in cluttered environments requires balancing formation maintenance, obstacle avoidance, inter-UAV safety, and per-step computational cost. This paper proposes a multilayer perceptron (MLP) surrogate for high-level objective-weight selection in a modified multi-objective pigeon-inspired optimization (modified MPIO) distributed controller. The proposed MLP surrogate learns the state-to-weight mapping of the online search and directly predicts the two-dimensional objective-weight vector, while the original flocking, gap-based obstacle-avoidance, and command generation rules are retained unchanged. The surrogate is trained from teacher-generated weight labels using randomized scenes, DAgger-based state aggregation, and risk-weighted supervision. On a fixed closed-loop benchmark, the proposed controller increases the true collision free rate from 48.00% to 86.89% and the safe success rate from 38.67% to 74.22% relative to modified MPIO, while reducing the mean per-step decision latency for the whole swarm from 8494.70 ms to 0.92 ms. The improvement is most pronounced in safety-related and runtime metrics, while the formation-related gain is comparatively modest. Ablation results show that the final benchmark performance is not explained by DAgger or risk weighting alone, and that the medium-sized surrogate provides the best safety-latency tradeoff among the tested network architectures. A qualitative AirSim case study further indicates that the same high-level surrogate controller can be executed in a higher-fidelity asynchronous multirotor simulator.

Sensors doi: 10.3390/s26113399

Authors: Hirotaka Inoue Tsukasa Komura Ibuki Hashimoto

Deep learning models, particularly convolutional neural networks (CNNs), have demonstrated exceptional classification performance across various practical applications. However, their training time scales significantly with network depth, rendering them suboptimal for resource-constrained environments. As a practical alternative, multiple classifier systems (MCSs) based on self-generating neural trees provide faster training and lower computational overhead. In this study, we propose the Self-Organizing Neural Grove (SONG), an ensemble learning model featuring a novel pruning technique designed to optimize classification efficiency. We evaluate SONG’s performance on a Raspberry Pi 3, a standard edge computing platform. Through comparative experiments against an unpruned MCS, a C4.5-based MCS, and the k-nearest neighbors (k-NN) algorithm, we demonstrate that SONG achieves superior classification accuracy while substantially reducing both computation time and memory footprint. These advantages are consistent across benchmark datasets and real-world cybersecurity tasks, underscoring the high suitability of SONG for edge computing applications.

Sensors doi: 10.3390/s26113397

Authors: Mohamed Amin Gader Sourour Karmani Ridha Djemal Carlos Valderrama Sakuyama

Heart failure (HF) remains a major global health challenge, necessitating accurate yet accessible diagnostic tools. While the left ventricular ejection fraction (LVEF) is the primary metric for classifying HF into preserved (HFpEF), mid-range (HFmrEF), and reduced (HFrEF) phenotypes, conventional imaging modalities such as echocardiography are resource intensive. In contrast, the electrocardiogram (ECG) offers a low-cost, non-invasive alternative for continuous cardiac assessment. This paper proposes a multi-algorithm artificial intelligence (AI) framework for automated HF phenotype classification using high-resolution ECG signals from 303 patients with chronic heart failure from the MUSIC cohort. After preprocessing (normalization, bandpass filtering), we employed a hybrid approach combining the Pan–Tompkins algorithm for robust R-peak detection with the NeuroKit2 toolbox for the precise delineation of P, Q, S, and T waves. ECG recordings were then segmented using an adaptive beat-centric windowing strategy. From the segmented beats, we extracted a comprehensive set of temporal, morphological, and energy-based features, including RR, QRS, and QT intervals, along with P-wave, QRS-complex, and T-wave energies. These features were used to train and evaluate several ensemble machine learning models—Random Forest, XGBoost, CatBoost, LightGBM, and a stacking classifier—using a stratified 70–15–15 train–validation–test split with 5-fold cross-validation. The LightGBM model achieved the highest performance with a test accuracy of 98.45%, an AUC of 0.9989, and a macro F1-score of 0.9804, outperforming other ensembles and the stacking classifier. The results demonstrate that an AI-driven analysis of ECG-derived morpho-energy features can serve as a reliable, non-invasive screening tool for the accurate and early discrimination of HF phenotypes, potentially supporting clinical decision making and improving patient management in resource-limited settings.

Sensors doi: 10.3390/s26113396

Authors: Zhiyuan Liu Wenxiu Wan Ziru Yu Zhiqie Jiang Xiangyang Yu Youliang Zhang Shengkang Luo Yuchen Guo Ke Chen

Three-dimensional (3D) hyperspectral point clouds provide both surface geometry and spectral information, offering a promising tool for close-range surface characterization. However, reliable reflectance-related spectral measurement on complex surfaces remains challenging because camera-recorded spectral signals are strongly affected by non-uniform illumination, surface geometry, and the spectral response of the imaging system, while existing correction methods are often limited by Lambertian assumptions and narrow spectral capacity. In this work, we present a close-range 3D hyperspectral measurement framework with geometry-aware spectral correction that integrates a structured-light 3D measurement module with a hyperspectral imaging module. The system enables the acquisition of fused 3D hyperspectral data with a sphere-fitting RMS residual below 40 μm and a spectral resolution of 7 nm. To improve spectral correction on geometrically complex surfaces, we propose a physics-guided spectral correction model, termed 3D light-field spectral correction (3D-LFSC), which is inspired by the geometric dependence described by the bidirectional reflectance distribution function (BRDF) and uses measurable geometric information to model geometry-dependent spectral variation. Because the system adopts a cross-polarized illumination–detection configuration, the corrected spectra should be interpreted as diffuse-dominant apparent reflectance estimates under the fixed system configuration, rather than complete surface reflectance. Experiments on surfaces with different geometries and reflectance properties show that the proposed method improves spectral consistency by more than 10% compared with existing methods. The framework also demonstrates applicability to chromaticity-related analysis on facial surfaces, indicating its potential for close-range spectral measurement of complex biological surfaces.

Sensors doi: 10.3390/s26113395

Authors: Abdullah Abdulhameed Yaqub Mahnashi

The contamination of aquatic environments, including treated and drinking water, by microplastics poses a significant threat to ecosystems and human health. Current detection methods often rely on slow laboratory-based tests and offline analysis, which do not support real-time monitoring. This paper presents a novel focusing and concentrating device designed to enhance the real-time detection of microplastics in flowing water. The device utilizes an electrokinetic manipulation mechanism to focus microplastics toward the center of the water flow inside a pipe or fluid channel. A set of 3D rectangular electrodes, with dimensions of 5 mm × 2.5 mm × 1 mm, are arranged circumferentially and longitudinally along the inner perimeter of the fluid channel to generate an intense, non-uniform electric field. Simulation results indicate that microplastics near the channel wall experience a repulsive force on the order of 10−16 to 10−10 N toward the channel center. The applied signal amplitude and the physical properties of the microplastics strongly influence this repulsive force. The trajectories and output concentration of microplastics are investigated under varied conditions. A Voltage of approximately 25 V and a flow rate of 0.05 m/s are found to be ideal for concentrating microplastics into a narrow particle stream, enabling more efficient downstream detection and analysis. Pre-concentrating microplastics in fluid channels prior to sensing is expected to increase sensor sensitivity and improve selectivity.

Sensors doi: 10.3390/s26113394

Authors: Zhichao Lin Xiuling Wang Yuyang Guo

Photovoltaic (PV) module faults, such as hotspots, diode short circuits, occlusions, and shadows, degrade power generation efficiency and safety. Existing manual inspection and single-modality methods show limited robustness under complex conditions, especially with illumination variations and weak thermal responses, while most deep learning approaches fail to exploit the complementarity of visible and infrared modalities. To address this issue, a dual-modality visible–infrared fusion framework based on YOLO11 is proposed, integrating a multi-scale pyramid pooling and dilated convolution module (MSPPD), a gradient-aware fusion module (GAFusion), and a dynamic convolution and element-wise scaling detection head (Detect-DEhead). GAFusion enhances cross-modal structural consistency and reduces feature misalignment and information loss during fusion by introducing gradient-aware feature interaction. Shape-IoU loss is employed to improve localization accuracy. The proposed method improves mean average precision (mAP)@0.5 from 86.7% to 88.1%, while reducing parameters, computational cost, and model size from 4.3 M to 3.7 M, 11.42 GFLOPs to 9.37 GFLOPs, and 9.1 MB to 7.9 MB, respectively. With Shape-IoU, mAP@0.5 reaches 88.4%, and recall increases from 78.5% to 84.9%. Experiments on the FLIR Thermal dataset achieve gains of 2.2%, 1.6%, and 2.7% in precision, recall, and mAP@0.5. The method achieves an effective trade-off between accuracy and efficiency for intelligent PV module inspection.

Sensors doi: 10.3390/s26113393

Authors: Mariola Gepfert Artur Gołaś Adam Maszczyk Adam Zając Anna Zwierzchowska

This study examined long-term neuromuscular and multidirectional speed development in elite youth badminton players and evaluated whether developmental stage influences adaptation trajectories during systematic training. Thirty athletes were monitored over 16 months with repeated assessments at five time points and stratified into Younger (8–14 years) and Older (15–22 years) developmental groups. A comprehensive test battery assessed explosive strength, reactive strength, musculotendinous stiffness, and badminton-specific multidirectional speed. Data acquisition was performed using a multi-sensor approach, including force-platform-based jump analysis, accelerometry-based systems, and electronic timing gates, enabling the objective, high-resolution, and repeatable monitoring of neuromuscular performance. Significant time effects were observed across all sensor-derived performance variables (p < 0.001), indicating robust improvements in speed, power, and neuromuscular efficiency. Adaptation trajectories were predominantly linear, with no evidence of performance plateauing. Although older athletes maintained higher absolute performance levels, Time × Group interactions were largely absent, demonstrating parallel improvement rates across developmental stages rather than a catch-up effect in younger players. Linear mixed models confirmed equivalent improvement slopes despite baseline differences, and adjustment for body mass attenuated but did not eliminate age-group differences in jump performance. Exploratory analyses revealed substantial inter-individual variability, identifying responder phenotypes independent of age. These findings indicate that systematically progressed training supports sustained, linear neuromuscular adaptation across youth badminton development and highlight the importance of long-term, individualized monitoring over age-based expectations of accelerated responsiveness.

Sensors doi: 10.3390/s26113391

Authors: Yuntae Ha JinBeom Kwon Saewan Kim Dong Geon Jung Daewoong Jung

The use of LIDAR sensors is rapidly increasing across various fields, including autonomous transportation, robotics, security systems, and biosensing. Among the core components of LIDAR systems, short-wave infrared (SWIR) sensors play a crucial role in detecting infrared light reflected from objects to recognize the surrounding environment and humans. Various types of SWIR sensors have been reported, with growing demand for those capable of detecting eye-safe infrared wavelengths above 1400 nm. In particular, quantum dot (QD)-based SWIR sensors are attracting attention due to their tunable wavelength range within the eye-safe region, narrow full width at half maximum (FWHM), and selective detection with minimal interference from other infrared wavelengths. Moreover, QD-based SWIR photodetectors can be synthesized and fabricated via solution-based methods, offering advantages such as low cost and ease of fabrication. However, the long organic ligands typically present on QDs exhibit insulating properties, limiting the sensitivity and stability of the photodetectors. To address this issue, organic ligands can be replaced with short inorganic ligands possessing superior electrical conductivity. In this study, the organic ligands of synthesized PbS QDs were replaced with the inorganic ligand methylammonium lead iodide (MAPI) in solution, and a SWIR photodetector was fabricated. The MAPI-capped PbS QD-based photodetector exhibited remarkable external quantum efficiency (EQE) of 62%, a responsivity of 0.73 A/W, and a detectivity of 2.26 × 1011 Jones within the 1400–1500 nm wavelength range.

Sensors doi: 10.3390/s26113392

Authors: Jiashun Dai Hong Lu Yukuo Guo Hao Xue Jiangnuo Mei Qiong Wang

With the increasing complexity of operating conditions and the trend toward structural compactness in generators, the unbalanced electromagnetic force induced by air-gap eccentricity has become a critical factor affecting rotor dynamic behavior and operational reliability. To address the strong coupling and modeling challenges among the electromagnetic field, mechanical force field, and lubrication flow field under eccentric conditions, this study proposes a multi-physics coupled modeling approach that integrates electromagnetic, structural, and fluid dynamic interactions. Based on the spatial pose characteristics of the rotor under eccentric conditions, a three-dimensional mathematical model of the air-gap length is established, and an analytical expression for the lubricating oil film thickness distribution is derived. This framework enables the coupled solution of unbalanced electromagnetic force, hydrodynamic oil film supporting force, and rotor dynamic response. A 60 kW-rated diesel generator was selected as the research object for both numerical simulations and experimental investigations. The numerical results indicate that when the load power increases from 0 kW to 60 kW, the displacement amplitude of the rotor in the y-direction increases by approximately 155%, demonstrating a significant enhancement of transverse vibration intensity under increasing unbalanced electromagnetic excitation. Comparison between experimental and numerical results shows good agreement in both variation trends and amplitude levels, with a maximum relative error of 4.07%, thereby validating the accuracy and reliability of the proposed electromagnetic–structural–fluid coupled model for predicting rotor dynamic response in generators.

Sensors doi: 10.3390/s26113390

Authors: Xin Li Wei Shen Jian Li Zhongsheng Zhai Xuhong Guan Zili Lei

Shape-from-Focus (SFF) is attractive for microscopic three-dimensional measurement, but high dynamic range (HDR) surfaces and weak-textured surfaces distort the focus curve through saturation, spurious peaks, and low signal-to-noise ratios. These distortions violate the unimodal assumption used by Gaussian peak localization and limit post-processing-only correction. This paper proposes a physics-guided distortion-aware SFF pipeline for opaque single-surface targets. The Distortion-Aware Focal Depth Regression Network (DAFDR-Net) learns from synthetic focus-curve distortions and uses Channel-wise Feature Attention (CFA) and Soft Peak Localization to reweight distortion-sensitive temporal-response features while preserving a peak-localization prior. Its foreground validity output is further used for confidence-guided adaptive smoothing. On an HDR free-form surface dataset, the proposed pipeline reduces RMSE by 36.5% relative to an MRF optimization method and compresses the 99th-percentile absolute error from 0.181 to 0.033. On weak-textured monocrystalline silicon wafer data, it reduces flat-region depth standard deviation by 51.3%.

Sensors doi: 10.3390/s26113389

Authors: Xiaoqing Chen Tao Yang Bowen Zhang Ling Zhang

Sensorless permanent magnet synchronous motor (PMSM) drives rely on accurate rotor electrical angle and speed estimation, vulnerable to noisy currents, quantization, and sensor biases. Fixed-bandwidth phase-locked loops (PLLs) entail an intrinsic trade-off between fast transient tracking and high-frequency noise rejection. This paper proposes an adaptive PLL based on linear active disturbance rejection control (LADRC), where a virtual coordinate formulation treats electrical-angle mismatch as a lumped disturbance estimated online by a linear extended state observer (LESO). The observer bandwidth dynamically adapts to the LESO innovation. To optimize performance, adaptive-law parameters are tuned offline via success-history adaptive differential evolution with linear population size reduction (L-SHADE). Comparative simulations against a proportional-integral PLL indicate substantially improved robustness to measurement noise, analog-to-digital quantization, and current-sensor DC offset. Specifically, the speed root-mean-square error decreases from 68.9r/min to 20.7r/min under 0.15A additive noise, and from 1.55r/min to 0.48r/min under 12-bit quantization at 200r/min. These enhancements reduce reliance on high-precision sensing hardware, offering a practical solution for low-cost, highly reliable motor control in complex industrial environments.

Sensors doi: 10.3390/s26113388

Authors: Ang Li Wei Lu

High-quality cardiopulmonary resuscitation (CPR) aims at saving lives in time-critical emergencies, which requires correct compression rate, depth, and hand placement. However, due to the high cost and environmental constraints of sensor-equipped manikins or dedicated hardware, it is unrealistic to deploy these devices in ordinary training settings. For monocular vision-based methods, estimating compression depth without direct depth signals and tracking hands under severe overlap are difficult. To address these problems, this paper proposes VISIOCPR, a monocular vision-based CPR training system with human-computer collaborative feedback, which provides quantitative CPR coaching using only a standard RGB camera. To address the inherent visual constraints, the system integrates a tiered compression-point detector that maintains robust tracking continuity despite severe hand overlap and motion blur. Furthermore, it recovers accurate metric depth without attached markers through a fused calibration scheme, which combines an empirical baseline, a reference-object measurement, and visible body proportions. A randomized controlled study (n=40) showed that participants trained with VISIOCPR achieved higher simultaneous compliance and reached competency faster than the control group under the tested setting.

Sensors doi: 10.3390/s26113387

Authors: Nick Barua Masahito Hitosugi

Pedestrian–vehicle collisions produce a rich kinematic record that is entirely lost by the time a forensic investigation begins. Recovering this record constitutes a state-estimation problem. This paper presents a Phase 1 design for a multimodal sensor fusion and signal-processing framework utilising 128-channel LiDAR, 1080p NIR stereo cameras, and a 2 kHz IMU, all fused via Kalman filtering and Savitzky–Golay polynomial differentiation. The framework is evaluated through Monte Carlo uncertainty propagation and sensitivity analysis applied to a constructed simulation scenario; no real clinical or forensic data are used in this Phase 1 report. Under simulated conditions with throw-distance measurement uncertainty of ±0.5 m, velocity reconstruction shows an estimated propagated uncertainty of ±2.03 km/h under expanded simulation conditions with vehicle-coefficient variance activated. Sensitivity analysis indicates that a 10% noise spike in acceleration would theoretically amplify injury metrics by 26.9%, providing quantitative justification for noise-optimal pre-filtering. The bimodal kinematic–acoustic architecture is proposed as a physically interpretable foundation for collision reconstruction; its experimental performance awaits Phase 2–4 validation. A five-phase validation roadmap is presented, progressing from FEA simulation to independent multi-site replication before any forensic deployment is proposed.

Sensors doi: 10.3390/s26113386

Authors: Rohit Srivastava Ved Prakash Bhardwaj Mohamed Tahar Ben Othman Mukesh Pushkarna Anushree Arushi Mangla Mohit Bajaj Ateeq Ur Rehman Muhammad Shafiq Habib Hamam

The Journal retracts the article “Match-Level Fusion of Finger-Knuckle Print and Iris for Human Identity Validation Using Neuro-Fuzzy Classifier” [...]

Sensors doi: 10.3390/s26113384

Authors: Yao Liu Gui Wang Daguo Yu Huifu Du

To address the issues of uneven accuracy across the entire hole depth and profile distortion caused by multi-source errors in spectral confocal deep-hole measurement, this paper proposes a measurement method involving global calibration using a stepped ring gauge and hierarchical compensation for multi-source errors. By classifying core measurement errors into three categories—geometric deviation, structural error, and dynamic process error—according to their propagation laws, this paper establishes a progressive comprehensive compensation system comprising “geometric calibration–structural correction–dynamic filtering”. Specifically, using a stepped ring gauge as the reference, the system’s intrinsic geometric parameters are identified via the Levenberg–Marquardt (LM) algorithm; structural errors introduced by the deflection of components due to self-weight are quantitatively corrected based on a statics model; periodic harmonic errors are sequentially separated; random noise is effectively suppressed by combining least-squares harmonic fitting with adaptive wavelet threshold filtering. Experimental results demonstrate that this method can limit the maximum absolute deviation in the inner diameter measurement of standard ring gauges to within 0.2 μm, stabilizing the measurement repeatability over the entire depth of deep-hole workpieces with length-to-diameter ratios exceeding 30:1 to within 0.8–1.6 μm, with an expanded uncertainty of U = 3.8 μm (k = 2). This method enables the precise reconstruction of deep-hole inner wall topography, providing a highly versatile technical foundation and implementation scheme for the high-precision non-destructive testing of deep holes with large length-to-diameter ratios.

Sensors doi: 10.3390/s26113385

Authors: Zhiqi Gao Ruyu Zuo Pingping Huang Wei Xu Weixian Tan Zhixia Wu

Conventional adaptive beamforming algorithms often suffer from significant performance degradation when steering vector mismatches and covariance matrix estimation errors occur. To address this problem, this paper proposes an adaptive robust beamforming algorithm based on an improved generalized linear combination (GLC) framework. The proposed method first applies singular spectrum analysis to the received signals to suppress noise components. A diagonal loading coefficient function related to the received signal snapshots is then constructed, and a generalized diagonally loaded covariance matrix is formed using the denoised data. Finally, by exploiting spatial integration and subspace projection within a predefined angular uncertainty set, the actual direction of arrival of the desired signal is accurately estimated, and the steering vector is corrected accordingly. Simulation results demonstrate that, compared with traditional SMI, LSMI, GLC and improved GLC algorithms, the proposed method achieves a 3–5 dB higher output signal-to-interference-plus-noise ratio (SINR) across the entire input signal-to-noise ratio (SNR) range under steering vector mismatch, and reaches an output SINR close to the optimal level with only 100 snapshots, exhibiting excellent robustness against steering vector mismatch and limited snapshot conditions.

Sensors doi: 10.3390/s26113383

Authors: Muhammad Arslan Muhammad Faran Majeed Rana Abu Bakar Jawad Khan Shafiq Hussain Youngmoon Lee Faheem Khan

The journal retracts the article, titled “CAVVPM: Challenge-Based Authentication and Verification of Vehicle Platooning at Motorway” [...]

Sensors doi: 10.3390/s26113382

Authors: Senming Zhong Chen Shu Liancai Shen Zhangjun Wu Minglong Xue Xiaojun Wang Weiwei Zhu

This review summarizes the research on the end effectors of agricultural harvesting robots (2010–2025) and extracts two core design principles. First of all, the selection of end effectors must follow the biological characteristics of fruits: rigid grippers are suitable for hard skinned and regular fruits; soft grippers can reduce the damage of fragile crops to a certain extent; suction cups are suitable for smooth, barrier free surfaces; the envelope type is suitable for soft and lossless picking scenes; the combined suction and grip design is more suitable for unstructured environments. Secondly, the separation mode should match the characteristics of the stem: motion separation (torsion/pull) is suitable for weak stems, while cutting is mainly used for hard stems. Unlike previous literature, this review provides a field deployability checklist (including dust/water proofing, cleanliness, maintenance, aging prevention, and aspiration prevention) to narrow the results of the laboratory and the real field environment. The three future directions of multimodal perception, variable stiffness driving and reinforcement learning are logically related to the analysis in this paper: multimodal perception optimizes the perception limit, variable stiffness solves the rigid–flexible trade-off, and reinforcement learning provides adaptive strategies for different crops. This framework can match the end effector design with the crop-specific field conditions.

Sensors doi: 10.3390/s26113381

Authors: Jialong Wang Jianli Wang Jiaxin Jing Canyang Zhao Lei Zhang

Traditional sliding mode control (SMC) lacks an active mechanism for redistributing energy among state channels during transient convergence, resulting in a rigid trade-off between response speed, overshoot suppression, and energy efficiency. This paper proposes a virtual state coupled SMC method that introduces a dynamic virtual state with bilinear product coupling x1x2 into the sliding surface. Unlike conventional virtual states that serve as static linear combinations or observer-based estimates, the proposed virtual state evolves dynamically and establishes an active energy exchange channel between the real and virtual state dynamics. Linearization and Lyapunov-based analyses prove local asymptotic stability of the closed-loop system. The coupling strength γ is shown to be decoupled from the linearized local eigenvalues and thus governs the energy–performance trade-off independently, while the condition c>γ/4 guarantees a non-vanishing domain of attraction. Simulations demonstrate that the proposed method achieves up to 53.2% control energy reduction under disturbance-free conditions compared with conventional SMC. Under persistent high-frequency disturbances, increasing γ reduces oscillations by 54.2% at a controllable energy cost of 45.7%. Systematic parameter selection guidelines are provided, and Monte Carlo simulations (500 trials, ±30% parameter perturbations) confirm 100% convergence. The proposed method offers an independently adjustable energy–performance trade-off mechanism suitable for sensor-based motion systems with stringent transient and energy requirements.

Sensors doi: 10.3390/s26113380

Authors: Jinmei Li Ming Zhao Quan Du Song Lu Shenglung Peng

Colonoscopy polyp segmentation is important for colorectal cancer screening, yet it remains challenging because polyps exhibit large morphological variation, weak lesion–background contrast, blurred boundaries, and severe foreground–background imbalance. To address these issues, this paper presents MCM-UNet++, a hybrid U-Net++-based segmentation framework that combines three targeted enhancements. First, a Multi-Axis Transformer Block (MATransformerBlock) is incorporated into convolutional feature blocks to model long-range horizontal and vertical dependencies with lower complexity than dense global self-attention. Second, a Cross-Channel Mixing (CCM) module is used in nested skip fusion paths to recalibrate the channel and spatial responses and reduce redundant feature transmissions. Third, a Multi-Objective Adaptive Loss (MOALoss) combines focal, Dice, and boundary-aware terms with learnable weights to improve supervision for small regions and ambiguous boundaries. Experiments on four public polyp segmentation datasets (Kvasir-SEG, CVC-ClinicDB, CVC-ColonDB, and ETIS-Larib) show competitive performance against the selected baseline methods, with Dice/IoU scores of 0.9563/0.9278 on Kvasir-SEG and 0.8593/0.7896 on CVC-ColonDB. These results indicate that the proposed components can improve benchmark-level polyp segmentation performance, while broader validation is still required before clinical deployment.

Sensors doi: 10.3390/s26113379

Authors: Ehtisham Muhammad Xiao-Feng Sun Annum Zia Ran Sun Sihai Hu

The growing demand for reliable, real-time detection of ammonia (NH3) has accelerated the development of chemiresistive gas sensors, while conventional semiconductors employed as sensing materials in chemiresistive sensors remain constrained by limited selectivity and high operating temperatures (typically 200–400 °C). Among the emerging porous materials, metal–organic frameworks (MOFs) have attracted significant attention as room-temperature NH3 sensing materials owing to their structural tunability, enabling precise control over pore chemistry, functionality, and metal centers. However, a comprehensive study specifically focused on MOF-based chemiresistive NH3 sensors operating at room temperature remains limited. This review critically targets the investigation of pristine MOFs, conductive MOFs, and MOF-based composites for NH3 sensing, with an emphasis on sensing mechanisms, structure–property–performance relationships, stability, selectivity, and environmental effects. Furthermore, rational design strategies and prospects are discussed to provide guidelines for the development of next-generation high-performance room-temperature NH3 chemiresistive sensors.

Sensors doi: 10.3390/s26113378

Authors: Louise C. Burgess Matthew Armstrong Louise Beresford Andrew J. Callaway

Background: The use of wearable sensors to measure and monitor heart rate has exponentially grown in recent years, representing an inexpensive, time-efficient, and non-invasive method to assess the status of cardiovascular fitness and the autonomic nervous system. Validating new devices against a criterion standard, such as electrocardiography (ECG), is essential to ensure their accuracy and reliability. This study examined the accuracy and validity of the Prevayl heart rate monitor against 3-lead ECG. Methods: Twenty-six healthy adults (15 female, mean age 32.0 ± 10.4 years) completed a 16-min, incremental running test on a treadmill. Heart rate data were recorded simultaneously throughout the test via ECG and the Prevayl wearable and compared retrospectively. Beat count error (%), mean heart rate absolute error (beats per minute (bpm)), and percentage error (bpm) were calculated. In addition, a Bland–Altman analysis and Pearson’s correlation coefficient were conducted to assess agreement and correlation, respectively. Results: The Prevayl device demonstrated a median beat count agreement of 100.5% with ECG (range: 98.6–104.4%; Npart = 26). Strong correlations were observed between ECG and Prevayl for both raw beat count (r = 0.94, p < 0.01) and heart rate (beats per minute (bpm)) from ECG and the Prevayl algorithm (r = 0.96, p < 0.01). Across running speeds (0–12 kph), a strong correlation was found between raw beat count from ECG and Prevayl (r = 0.82–0.89, p < 0.01) and between bpm from ECG and Prevayl (r = 0.86–0.93, p < 0.01). Bland–Altman plots demonstrated negligible systematic bias. Conclusions: The Prevayl system provides valid measurements when compared to ECG during incremental running. This is demonstrated through strong correlations to ECG heart rate data at different speeds and with different analysis methods, supporting its use for monitoring cardiovascular responses during exercise.

Sensors doi: 10.3390/s26113377

Authors: David O. Briceño González Julian Sierra-Perez Maribel Anaya Vejar Diego Tibaduiza Burgos

Structural damage classification in composite UAV wings is a key challenge in Structural Health Monitoring (SHM), particularly under barely visible impact damage conditions. Fiber Bragg Grating (FBG) sensor networks provide high-resolution strain data; however, systematic experimental benchmarking of lightweight neural architectures trained on real FBG datasets remains limited, especially under sensor degradation scenarios. This work presents a four-phase benchmarking study of Multilayer Perceptron (MLP) configurations using strain measurements from a composite UAV wing instrumented with 32 FBG sensors across five damage states and 210 loading experiments. The framework evaluates optimization strategies, hyperparameter sensitivity, architectural depth, and robustness under controlled sensor dropout, Gaussian noise, and wavelength drift perturbations. Results indicate that compact architectures with progressive dimensional reduction (256–128–64) trained using adaptive optimizers (AdamW and Nadam) achieve the best balance between macro-F1 performance (up to 0.85 during validation), stability, and computational efficiency. Robustness analysis shows gradual performance degradation under sensor loss, suggesting distributed strain-field learning. These findings provide practical guidelines for selecting computationally efficient and robust neural models for deployable FBG-based SHM systems in aerospace applications.

Sensors doi: 10.3390/s26113376

Authors: Wei Bai Yan Zhao Yanni Ju Jing Gan Linheng Li

Investigating the fundamental link between microscopic vehicular motion parameters and macroscopic traffic flow states is pivotal for advancing refined traffic state estimation research and propelling the progression of Intelligent Transportation Systems. In this paper, a basic Transformer model has been optimized and extended by incorporating embedding and pooling layers, and the model’s hyperparameters have been finely tuned through random search cross-validation. The creation of the Generalized Optimized Transformer (GOT) model ensued, where the multi-head attention mechanism adeptly encapsulates all spatiotemporal dynamics inherent in traffic data. Various benchmark models such as LSTM, RNN, and Transformer were put to test, each demonstrating unique performances in managing different traffic flow states. Among them, the GOT model exhibited superior performance, adeptly handling lane-level traffic state estimation tasks derived from microscopic vehicle trajectory data. In conclusion, this research elucidates the intricate and mutable mapping relationship between microscopic vehicular motion parameters and traffic flow states, proficiently executing lane-level traffic state estimation grounded on microscopic trajectory data. This paper is anticipated to provide fresh insights into the understanding of the complex relationship between microscopic vehicular motion parameters and traffic flow states.

Sensors doi: 10.3390/s26113374

Authors: Lufan Xie Lijing Zhang Fan Yang Mengchen Lin Jiadong Wang Shengxiang Cao Chenlong Zhang Di Liu Mingzhong Pan Jin Yang

Traditional hyperspectral cameras transmit full data cubes to host computers, creating severe bandwidth and storage bottlenecks that impede real-time analysis. We present a Zynq-7035-based intelligent camera using hardware–software codesign to enable on-board processing and transmit only actionable results. This intelligent camera is designed for high-throughput edge-sensing tasks, prioritizing rapid detection and information extraction over exhaustive raw data acquisition. The processing system (PS) handles command scheduling while the programmable logic (PL) implements a row-parallel pipeline for image acquisition, preprocessing, and spectral matching; all modules are decoupled through a unified DDR3 interface to support flexible algorithm integration. Push-broom experiments on leaf samples demonstrate Euclidean distance-based spectral matching executed entirely within the camera. Raw data and classification maps are uploaded via User Datagram Protocol (UDP). Results confirm accurate identification of diseased regions with two orders of magnitude data reduction, validating the architecture for real-time hyperspectral processing.

Sensors doi: 10.3390/s26113375

Authors: Jiongyi You Zhicheng Lin Baifa Zhang

Vertical ground reaction force is an important parameter for describing the developmental characteristics of young children’s vertical jumping. However, its application in large-scale physical fitness monitoring and routine teaching practice is greatly limited. Previous studies have used OpenCap to estimate vertical ground reaction force during adult jumping tasks and have provided preliminary validation, but its effectiveness in young children remains unclear. To examine the correlation and agreement of vertical ground reaction force (GRF) estimated by the OpenCap markerless motion capture system during young children’s vertical jumping and to explore the characteristics of vertical GRF estimated by OpenCap during the vertical jump. Kinematic and kinetic data during vertical jumping were synchronously collected from 16 young children using the OpenCap markerless motion capture system and a three-dimensional force platform, with each child completing three trials. Kinematic data were acquired using the OpenCap markerless motion capture system, and the vertical acceleration of the whole-body center of mass was calculated to estimate vertical GRF based on Newton’s second law. Pearson linear correlation analysis and Bland–Altman analysis were used to examine the differences in characteristics between the estimated vertical GRF and the measured vertical GRF. The vertical GRF characteristics estimated by OpenCap showed moderate-to-high correlations with the measured values. Specifically, the time and mean impulse during the push-off phase, flight phase, and landing stabilization phase were highly correlated (r > 0.85), while the peak force and mean force during the push-off phase showed moderate-to-high correlations (r > 0.7). Bland–Altman analysis showed that the bias in time and impulse during the vertical jump was less than 15%, indicating relatively high agreement; however, the bias in peak force during the landing phase exceeded 40%, indicating weak agreement. These results suggest that the OpenCap markerless motion capture system can effectively estimate vertical GRF characteristics during young children’s vertical jumping, with the best performance observed for vertical GRF variables in the push-off phase. The method used in this study may be applied to obtain vertical GRF during young children’s vertical jumping in non-laboratory settings and to assist in evaluating the developmental level of young children’s vertical jump performance. Nevertheless, OpenCap-derived rapid impact variables, particularly landing peak force, should be interpreted with caution.

Sensors doi: 10.3390/s26113373

Authors: Yi Tang Ziyi Yang Zhoucong Xu You Zhou Hui Wang

Automated pavement distress detection (PDD) is critical for the structural health monitoring (SHM) of transportation infrastructure, yet existing methods struggle with real-time multi-target detection under resource constraints. In this paper, YOLOv8-PDD was constructed based on YOLOv8 by introducing the large separable kernel attention (LSKA) mechanism module into the Spatial Pyramid Pooling—Fast (SPPF) module, replacing Complete-IoU (CIoU) loss with Distance-IoU (DIOU) loss as the loss function, and adopting Soft-Non-Maximum Suppression (NMS) to replace the original NMS algorithm. The proposed YOLOv8-PDD achieved 78.3% mean average precision with intersection over union above 0.5 (mAP@0.5 +8.1%) with a minimal complexity increase of +0.2 GFLOPs compared to the baseline YOLOv8n model. While incurring a negligible increase in latency (+0.09 ms), YOLOv8-PDD significantly outperforms YOLOv8n in detection accuracy (mAP@0.5 +8.1%), offering a superior accuracy–efficiency trade-off for real-time applications. YOLOv8-PDD performed well in detecting all categories, with AP values above 75% except for transverse crack and strip patch. Significant improvements in pothole detection AP@0.5 (+22.1%) and strip patch detection AP@0.5 (+17.7%) indicate superior small target and complex background adaptability. Our model achieved a detection efficiency of 68 frames per second (FPS) on consumer-grade CPUs (OpenVINO-optimized), outperforming 10 models (e.g., YOLOv5n and RTDETR-l) in accuracy–speed balance.

Sensors doi: 10.3390/s26113369

Authors: Yichen Qian Taiming Kang Shengduo Zhang Chaoneng Li Xiaolong Wang Shuxu Zhao

Traffic flow forecasting remains challenging because raw traffic flow observations often contain mixed temporal patterns, including slowly varying trends and fast local fluctuations. To address this issue, this paper proposes a Multivariate Empirical Mode Decomposition (MEMD)-guided dual-branch recurrent framework for multistep point forecasting. Specifically, MEMD is used as an alignment-preserving multivariate decomposition mechanism to obtain frequency-aligned components, which are then reconstructed into low-frequency trend and high-frequency residual components. The trend component is modeled by a Long Short-Term Memory (LSTM) branch to capture smooth long-term evolution, while the residual component is learned by a Bidirectional Gated Recurrent Unit (Bi-GRU) branch to characterize short-term oscillatory dynamics. A lightweight fusion head is then used to integrate the two branch-specific representations for final prediction. Experiments on PeMS04 and PeMS08, two traffic datasets derived from the California Department of Transportation Performance Measurement System, show that the proposed method achieves competitive performance across mean absolute error (MAE), root mean square error (RMSE), and mean absolute percentage error (MAPE), reaching 19.67/31.59/12.95% on PeMS04 and 15.51/24.43/9.86% on PeMS08. Compared with representative recent baselines, the proposed method achieves competitive results, with relative gains reaching 5.89% on PeMS04 and 5.35% on PeMS08 in selected metric-wise comparisons. These results indicate that MEMD-guided trend–residual representation learning can improve multistep traffic flow forecasting.

Sensors doi: 10.3390/s26113370

Authors: Víctor González Juan Álvaro Fernández Patricia Arroyo Jesús Lozano

A portable electronic nose based on modern digital metal oxide (MOX) gas sensors and programmable temperature modulation was developed and validated. The system integrates four modern commercially available MOX sensors capable of generating temperature-dependent odor fingerprints and multidimensional sensor responses compared with conventional fixed-temperature operation. The performance of the device was assessed in terms of sensor stability, repeatability, and pattern-recognition capability under long-term operation. As a proof of concept, the electronic nose was applied to the discrimination of Extra Virgin Olive Oil and pomace oil. Repeatability analysis using the Root Mean Squared Error (RMSE) demonstrated stable responses across one month of measurements. Temperature-modulated signals were processed using Principal Component Analysis (PCA) and classified with k-Nearest Neighbors (KNNs) and Multilayer Perceptrons (MLPs), achieving 100% accuracy after selecting the most repeatable sensor. These results highlight the robustness and analytical potential of temperature-modulated digital MOX sensors and demonstrate the feasibility of a compact and highly reproducible electronic-nose platform suitable for complex odor-analysis tasks in real-world applications.

Sensors doi: 10.3390/s26113372

Authors: Zuzhen Huang Aifang Liu Rui Zhang Long Li Jinjian Cai

The squint synthetic aperture radar (SAR) offers flexible beam pointing control and a wider range of applications compared to the side-looking SAR. Unlike the latter, ground moving targets exhibit shifts in both along-track and cross-track directions in squint SAR systems. To address this issue, a two-dimensional relocation method for moving targets is proposed in this paper. Firstly, the shift characteristics of moving targets in squint SAR systems are analyzed, revealing that the two-dimensional location shifts are correlated with both the target’s radial velocity and its imaging location. The proposed algorithm initially performs clutter suppression on the SAR imagery and estimates the radial velocity of the moving target. The two-dimensional location information is then derived by solving a set of joint equations. Finally, some numerical experiments are provided to demonstrate the effectiveness of the proposed method in the squint SAR system.

Sensors doi: 10.3390/s26113371

Authors: Hai-Bo Wang Zhen-Zheng Zhang Qing Liu Hang-Bo Lu Jian-Hui Jiang Ru-Qin Yu Hao Tang

In this paper, recent advances (2016–2026) in bacterial separation and enrichment from blood for diagnosis of bloodstream infection (BSIs) through pathogen identification and antimicrobial susceptibility testing (AST) are reviewed. The review centers on sample processing as an indispensable front-end of biosensor and lab-on-chip platforms, since most sensors cannot operate directly in whole blood. Efficient separation and enrichment concentrate extremely low bacterial burdens, remove blood components that interfere with detection, and deliver bacteria in a sensor-compatible format; consequently, diagnostic sensitivity, specificity, turnaround time, and robustness are strongly determined by this step. We first summarize the clinical impact of BSIs and the value of rapid AST for guiding timely, targeted therapy, emphasizing that efficient bacterial isolation from blood is a prerequisite for accurate testing. We then discuss key challenges and recent progress in bacterial separation and enrichment from blood with major approaches, including filtration, centrifugation, functionalized magnetic beads, and microfluidic technologies. These strategies serve as core building blocks that interface with downstream identification and AST methods, supporting integrated biosensors and point-of-care devices. Finally, we outline future directions of bacterial separation and enrichment approaches to improve recovery, purity, integration, standardization, and overall diagnostic performance for BSI workflows.

Sensors doi: 10.3390/s26113368

Authors: Jan Kaczyński Paweł Rzucidło

This paper explores the feasibility of using smartphones as emergency flight data indication systems in light aircraft. The presented solution may be applied in potential situations such as failures of the vacuum system or the gyroscopes driving analog instruments, as well as electrical power failures in aircraft equipped with digital avionics. Such failures may lead to the loss of essential flight information, significantly increasing pilot workload and conceivably compromising flight safety. The analysis was based on simulations conducted in a computational environment utilizing a custom-developed model. An experimental measurement flight using the MP-02A “Czajka” aircraft was conducted to collect real flight data for integration into a computational model. During the test flight, the aircraft was deliberately maneuvered into various attitudes and flight conditions to evaluate the model’s performance across the widest possible range of operating states. A smartphone mounted in the cockpit recorded sensor data, including accelerometer, gyroscope, magnetometer, and GPS information. The results demonstrated that key flight parameters can be accurately determined using only data recorded by a smartphone. For example, the determined pitch angle values during stall maneuvers deviate from the reference values by no more than 5°. The proposed solution shows significant potential for further development and practical implementation as a supplementary system to assist pilots during in-flight emergencies.

Sensors doi: 10.3390/s26113367

Authors: Xin Zhang Xuewei Xiang Hui Li Nengqing Liu Zhi Chen

Remaining useful life (RUL) prediction is a critical procedure to avoid catastrophic failure of shielding sleeves and prevent nuclear safety risks. Affected by the structural characteristics and service conditions of shielding sleeves, it is difficult to obtain sufficient full-life-cycle actual degradation data, which greatly restricts the training and application of data-driven prediction models. This paper proposes a remaining useful life prediction method for shielding sleeves based on data augmentation and a hybrid model. Firstly, starting from the physical failure mechanisms of two typical failure modes of shielding sleeves, namely bulging and wear. Secondly, based on the analytical models of typical failures of shielding sleeves, a degradation data augmentation method using Monte Carlo simulation is proposed to address the problem of missing full-life-cycle degradation data. Finally, a hybrid RUL prediction model for shielding sleeves based on Stacking ensemble learning is presented, which integrates the advantages of physical degradation models and deep learning methods. Experimental verification is carried out through multiple sets of degradation datasets with different failure modes. The root-mean-square error (RMSE) of the proposed prediction method can reach a minimum of 0.0058, and the mean absolute error (MAE) can reach a minimum of 0.0044. The prediction accuracy is superior to that of single models, which verifies the prediction performance and engineering applicability of the proposed method.

Sensors doi: 10.3390/s26113366

Authors: Yu Xu Xizheng Li Yinghao Song Yiqi He Junxiong Peng Wangling Mei Kun Zhang Yuyang Liu Yue Sun Xianjun Wu

Soil moisture content plays a crucial role in precision agriculture and geological hazard monitoring, driving the need for stable, reliable, and high-precision sensors. Capacitive soil moisture sensors based on Frequency Domain Reflectometry (FDR) are widely adopted due to their favorable measurement performance, yet their accuracy is highly susceptible to environmental interferences such as temperature, salinity (electrical conductivity), and soil type. This paper systematically reviews current calibration strategies addressing these three factors, classifying them into hardware-based compensation and software-based calibration (including conventional mathematical and machine learning models). Furthermore, it critically analyzes the trade-offs of these approaches in terms of robustness, scalability, and field applicability. To break through current technical limitations, this review argues that future research must prioritize the physical decoupling of multi-parameter interferences under extreme conditions. Additionally, to overcome the generalization crisis of current data-driven models, adaptive strategies utilizing techniques like transfer learning are essential. Finally, implementing Edge-AI on resource-constrained hardware is crucial for achieving calibration-free or real-time online calibration strategies, ensuring long-term measurement accuracy.

Sensors doi: 10.3390/s26113364

Authors: Kaiming Lu Zhe Liu Runsheng Zhang Qingyang Xia Ruoxuan Wang

Extensive research has been conducted on vehicle queuing and dissipation near signalized intersections. However, existing prediction methods for vehicle departure time primarily rely on assumptions of steady-state homogeneous traffic flow, utilizing shockwave theory and vehicle kinematic modeling. These methods encounter challenges in addressing traffic uncertainties during queue formation and dissipation, particularly in scenarios involving multiple lanes. This paper introduces a novel approach by leveraging unmanned aerial vehicle (UAV) trajectory data to construct fleet state features and proposes a prediction method for tail vehicle departure time based on Gaussian process regression. The objective of this method is to optimize the green light crossing time window and eco-driving trajectory for connected vehicles at signalized intersections. The findings reveal that the departure time of the tail vehicle within a specified distance adheres to a Gaussian process, demonstrating the applicability of Gaussian process regression for departure time prediction modeling. The effectiveness of the proposed method was validated using a field-measured dataset collected from three typical multi-lane signalized intersections. Notably, compared to four benchmark models (linear regression, decision trees, multilayer perceptron neural networks, and eXtreme Gradient Boosting—XGBoost), the mean absolute percentage error (MAPE) was reduced by an average of 5.146% on the test set under a random 70/30 split. Additionally, a robustness assessment demonstrates that the proposed model performs well, albeit slightly less effectively than the XGBoost model. We emphasize that the conclusions are drawn for the studied intersections; generalization to unseen intersections requires further validation with cross-site data.

Sensors doi: 10.3390/s26113365

Authors: Jingshan Jia Shuyang Duan Meina Wu

This paper presents a high-sensitivity and anti-interference optical intensity differential curvature sensor based on seven-core fiber (SCF), and its performance is verified through simulation. The sensor adopts a “single-mode fiber–tapered SCF” structure, using the light intensity ratio between the peripheral core and the central core for curvature demodulation. It enhances the sensitivity while effectively suppressing common-mode interference. The simulation results show that when the cone zone diameter of the tapered SCF is optimized to 30 µm, the mode coupling between the cores is significant, forming a strongly coupled super-mode transmission system. Based on the intensity differential principle, this sensor achieves excellent linear response within the curvature range of 0–10 m−1, with a sensitivity of −0.145/m−1. The sensor has a compact structure and simple fabrication process, providing new ideas and solutions to break through the technical bottlenecks of existing fiber curvature sensors, and has broad application prospects in engineering monitoring fields.

Sensors doi: 10.3390/s26113363

Authors: Tugce Keles Kubra Yildirim Dahiru Tanko Suat Tas Irem Tasci Burak Tasci Gulay Tasci Turker Tuncer Sengul Dogan

Electroencephalography (EEG) offers a non-invasive and cost-effective means of probing brain activity during language processing; however, prior EEG-based language studies have been limited by small datasets, a predominant focus on native-speaker or speech-unit recognition rather than direct language detection, evaluation on only a small number of experimental settings, and frequent reliance on computationally intensive deep learning models with limited interpretability. The proposed feature engineering models classifies EEG segments by language and task mode. The languages are Arabic and Turkish. The modes are reading and listening. In this study, a signal refers to one fixed-length multi-channel EEG segment (14 channels × 15 s at 128 Hz). A channel refers to one electrode time series within that segment. To address these gaps, we curated a new EEG language detection dataset from 346 participants (98 Arabic and 248 Turkish) recorded in reading and listening modes, yielding 6364 EEG segments. Using this dataset, we proposed DiagPat, an explainable feature engineering (XFE) model that extracts transition table-based features from both EEG channels and signals through diagonal pattern analysis. The model combines DiagPat feature extraction with iterative neighborhood component analysis (INCA) for feature selection, at algorithm-based k-nearest neighbors (tkNN) classifier for prediction, and the Directed Lobish (DLob) symbolic language for explainability. We evaluated the framework across nine classification cases covering language detection, mode detection, and mixed multi-class settings. The proposed DiagPat-driven XFE model achieved more than 90% accuracy in all cases, with accuracies ranging from 92.14% to 99.35%, and generated case-specific cortical connectome diagrams that supported the interpretable characterization of language- and mode-related brain activity. Subject-independent results were also reported using leave-one-subject-out cross-validation (LOSO CV), where LOSO accuracies ranged from 29.75% to 83.50%. Thus, the 10-fold CV results show segment-level performance, whereas the LOSO results show subject-level generalization. Balanced accuracy and macro-F1 are also reported. These findings indicate that DiagPat provides an accurate, lightweight, and explainable framework for EEG-based language detection.

Sensors doi: 10.3390/s26113362

Authors: Xin Liu Xuezhao Kang Liqun He Jianrui Zhang Huyan Ting Xiaojun Yu

Hydration level (HL) is a critical physiological indicator of human health and functional status, and accurate HL monitoring is essential for applications in healthcare, sports, and wellness assessment. However, existing methods are either invasive and inconvenient or noninvasive but limited by system complexity and insufficient accuracy. To address these limitations, this study proposes a methodological approach for noninvasive computerized HL estimation based on galvanic skin response (GSR) signals, termed the PSAML approach, which integrates principal component analysis (PCA), successive decomposition index (SDI), and machine learning (ML) classifiers. A representative GSR dataset was collected from three healthy subjects under dehydrated, normal, and overhydrated states in sitting, standing, and posture-independent scenarios. After preprocessing, including outlier removal, Butterworth filtering, and time-window segmentation, conventional time-domain features were extracted and compared with PCA- and SDI-based representations. Six ML algorithms were used for classification. The results show that the conventional feature method achieved a maximum accuracy of 63.97%, whereas PCA-based feature reduction significantly improved performance, with PCA+SVM, PCA+LR, and PCA+LDA achieving accuracies above 99% in most cases. SDI-based features also demonstrated strong performance with suitable classifiers under smaller time windows. These findings demonstrate that the proposed PSAML approach provides an accurate and efficient solution for wearable noninvasive HL monitoring.

Sensors doi: 10.3390/s26113361

Authors: Samuel Tremblay Philippe J. Renaud Shawn M. Robbins David J. Pearsall Philippe C. Dixon

In ice hockey, automatic activity detection using wearable sensors and machine learning could provide objective feedback to support coaches and players during performance evaluation. The primary objective was to assess the predictive ability of a deep learning model to recognize common ice hockey stick striking actions (passing, shooting) from inertial measurement unit sensors. This study implemented a fully connected convolutional neural network model to classify seven ice hockey-related technical actions (wrist shot, slap shot, backhand shot, one-timers, pass, other, and rest) using acceleration data via two setups: an all-sensor configuration (17 sensors) and a hands-only sensor configuration (2 sensors) in 43 elite players. Data were split into 80/20 train/test sets, with a five-fold cross-validation applied to the training data. The train/test split was repeated 10 times with different random splits to assess stability of results. The model achieved high classification accuracy, with the all-sensor model reaching an average F1 score of 95.0 ± 3.0% and the hands-only model achieving 93.5 ± 1.6%. These findings support the use of convolutional neural networks for automatic shooting action classification in ice hockey and highlight the feasibility of using minimal sensor configurations, such as sensor-integrated gloves, for real-world applications. This approach could further enhance training practices by providing objective performance metrics and allowing coaches to deliver data-driven feedback to players.

Sensors doi: 10.3390/s26113360

Authors: Gang Ming Xiao-Wei Ye Da Hang Jian-She Qin Jie Li

Unet, a deep learning architecture, has become one of the most widely used models for crack detection in the tunneling field. Although it performs well in overall crack image segmentation, it still has issues of limited feature expression capability and inaccurate segmentation. To address these problems, DTA-Unet was proposed based on dynamic convolution decomposition (DCD) and triple attention (TA). Firstly, the model used Unet as the baseline network and replaced traditional convolutions in the encoding-decoding process with DCD to enhance its feature extraction ability. Secondly, TA was combined with attention gate (AG) in the skip connections of the network, eliminating redundant information in spatial and channel dimensions to highlight the crack area. Finally, the proposed model was tested on crack datasets and compared with the conventional Unet model, image processing algorithms, and other deep neural network models in terms of detection performance on the datasets. The results show that it outperforms other advanced methods in crack detection performance. The proposed method is of significance to the maintenance of shield tunnel cracks.

Sensors doi: 10.3390/s26113359

Authors: Jinghao Cao Xiajun Liu Rui Xue

The closed-loop testing of autonomous driving systems critically depends on large-scale libraries of diverse and realistic 3D vehicle assets, yet current pipelines still rely on labor-intensive modeling or multi-view capture, making efficient construction a key bottleneck. To overcome this bottleneck and enable convenient, cost-effective 3D asset generation, we propose a semantic prior-guided framework for accurate and robust vehicle point cloud reconstruction from casually captured single-view photographs. Our framework is built on a diffusion backbone but is fundamentally driven by two forms of prior knowledge: First, geometric and appearance priors from camera-aware image features, masks, and distance-transform maps are projected onto the evolving point cloud, compensating for the severe information loss in single-view inputs. Second, we introduce distillation-style regulators—pretrained neural networks that encode vehicle type and model semantics; they act as teacher networks that impose high-level constraints on the generated point clouds, transferring rich semantic knowledge and effectively regularizing the learning process. With these priors, our model infers vehicle-specific semantics from limited observations and reconstructs high-quality 3D point cloud assets. On the 3DRealCar++ dataset, our method clearly surpasses state-of-the-art point cloud baselines in both F-score and Chamfer Distance.

Sensors doi: 10.3390/s26113358

Authors: Cristina Fornacelli Costanza Cucci Andrea Casini Maurizio Aceto Marcello Picollo

Color measurement and monitoring of chromatic changes over time play a key role in the study and conservation of historical materials. In this context, hyper-spectral imaging (HSI) offers spatially resolved spectral information that can be converted into colorimetric data, although its quantitative reliability under in situ conditions remains challenging. This study evaluates the colorimetric performance of a HSI system (Specim IQ) through comparison with a reference spectrocolorimeter (Konica-Minolta CM-700d), combining laboratory measurements on certified standards and in situ analyses on purple-dyed manuscripts. Colorimetric coordinates (CIELAB) and color differences (ΔE00) were used to assess accuracy, precision, and systematic deviations. Under controlled conditions, HSI showed good agreement with reference measurements, although systematic biases were observed. In situ applications revealed reduced accuracy (average ΔE00 ≈ 4.3) due to material heterogeneity and acquisition constraints. Despite these limitations, HSI preserved consistent relative chromatic relationships, enabling meaningful comparative analysis. Spatially resolved mapping of colorimetric parameters proved effective for visualizing chromatic variability, dye distribution, and degradation patterns. These results demonstrate that, while not fully reliable for absolute colorimetric assessment in situ, HSI represents a powerful tool for non-invasive, spatially resolved color analysis of complex historical materials.

Sensors doi: 10.3390/s26113356

Authors: Yuta Ogai Masaomi Sanekata

In recent years, wearable sensing technologies have been widely used for motion analysis in sports; however, in kendo, motion evaluation still largely relies on subjective assessment, and quantitative approaches remain limited. This study proposes an embedded inertial measurement unit (IMU)-based sensing system integrated into a bamboo sword (shinai) for the motion analysis of kendo swings. The system incorporates a compact IMU and a microcontroller within the shinai, enabling unobtrusive measurement under realistic training conditions without affecting usability. Using the acquired sensor data, motion-related acceleration components were extracted with orientation estimation using the error-state Kalman filter (ESKF) based on six-axis IMU data, followed by gravity compensation and feature extraction based on the peak characteristics of the swing motion. The experimental results show that experienced practitioners exhibited significantly higher peak acceleration (p = 0.002) and smaller peak width (p = 0.022) than novice practitioners, indicating sharper and more efficient motion. No significant differences were observed in the secondary peak ratio. These results demonstrate that the proposed system can quantitatively capture the motion characteristics of kendo swings and distinguish practitioners of different proficiency levels, which highlights its potential for objective motion analysis and training support in kendo.

Sensors doi: 10.3390/s26113357

Authors: Shuo Chen Jiahang Du Minsheng Liu Qiuyu Peng Jiayi Mi

Recent advancements in sensor technology have made in-situ crack assessment of structures feasible. To investigate the correlation between the ultrasonic amplitude and metal fatigue life, an aluminum compact tension (C(T)) specimen was fabricated to simulate fatigue damage in thin plate structures. An experimental investigation of fatigue crack propagation was performed, wherein the specimen experienced cyclic uniaxial tensile loading at constant amplitude. The crack propagation behavior was analyzed, and the relationship between crack length and the associated loading cycles was determined. Additionally, the evolution of the ultrasonic signal during crack propagation was investigated, and the quantitative dependence of the ultrasonic characteristic parameter on crack length was revealed. Finally, a model correlating ultrasonic characteristic parameters with loading cycles was developed, enabling fatigue life evaluation. The proposed method demonstrates significant potential for evaluating the fatigue life of thin plate structures.

Sensors doi: 10.3390/s26113355

Authors: Paras Miglani Aryan Garg Harshvardhan Singh Avinash Chandra Vijay Kumar Rajkishor Kumar

Vehicle-to-everything (V2X) communication systems impose stringent latency and reliability requirements that are difficult to satisfy in highly dynamic wireless environments. Although reconfigurable intelligent surfaces (RISs) and unmanned aerial vehicles (UAVs) have independently demonstrated potential in enhancing wireless coverage, most existing RIS–UAV frameworks rely on idealized assumptions such as perfect channel state information (CSI) and static user scenarios. In this paper, a multi-RIS-assisted UAV-enabled V2X communication framework is proposed that explicitly accounts for vehicular mobility, latency constraints, and mobility-induced CSI aging. Multiple RIS panels are cooperatively deployed to eliminate coverage blind spots and ensure link continuity in realistic V2X environments. A joint UAV mobility and RIS phase optimization approach is proposed under outdated CSI to improve link reliability. Additionally, a time-varying performance analysis is carried out for understanding the dynamic behavior of signal-to-noise ratio (SNR) and average bit error rate (ABER) for mobility-aware CSI aging. Simulation results demonstrate that the proposed framework reduces the ABER by approximately 75% compared to a conventional single-RIS system under outdated CSI at 20 dB SNR (1.07×10−1 vs. 4.32×10−1), while substantially suppressing outage intervals in high-mobility V2X scenarios (v=20 m/s, CSI delay τ=20 ms), confirming the effectiveness of cooperative multi-RIS assistance for safety-critical vehicular communications.

Sensors doi: 10.3390/s26113354

Authors: Jiuyang Yu Pan Liu Yaonan Dai Zelin Fu Hui Zhou Peiyan Yang Xiaotao Zheng

To address the detection challenges in X-ray weld seam images caused by weak contrast, slender structures, and multi-scale coexistence, we propose MGDR-YOLO, an industrially deployable detector with four coordinated designs. First, a MultiBackbone parallel heterogeneous backbone is designed to perform complementary direction–detail modeling and lightweight context modeling under a shared shallow stem, enhancing the joint representation of fine-grained features and global semantics. Second, Gated Attention Fusion Block (GAFB) is introduced to perform selective in-scale fusion via channel gating and local–global attention mechanisms, thereby suppressing channel redundancy and noise leakage induced by naive concatenation. Third, Directional Feature Convolution (DFConv) decouples standard 2D convolution into horizontal and vertical branches and fuses them using depthwise separable convolution, substantially reducing computational cost while preserving geometric alignment. Finally, Rep Shared Convolutional Detection Head (RSCD) improves detection head consistency and inference throughput through cross-scale shared convolutions and a training-to-deployment re-parameterization scheme. The experimental results show that MGDR-YOLO significantly outperforms YOLOv11n, increasing the mean average precision (mAP) from 92.9% to 95.2%. The performance gain is most pronounced for the LP class (slender and low-contrast defects), with an mAP improvement of 10.1 percentage points. Meanwhile, the proposed model achieves a 39.4% increase in frames per second (FPS) while reducing the number of parameters by 46.2%, demonstrating superior efficiency. These results indicate that MGDR-YOLO consistently improves the accuracy and robustness of X-ray weld defect detection while maintaining real-time performance, making it well suited for resource-constrained industrial online inspection scenarios.

Sensors doi: 10.3390/s26113353

Authors: Kexiang Li Ye Sun Chuang Li Tongzan Guo Hui Li

Preliminary estimation of lower-limb motor function is important in rehabilitation research, especially for biomechanical assessment of post-stroke hemiplegic gait. However, subject-specific musculoskeletal modeling in this population is challenging because standard maximum voluntary contraction (MVC) testing is often unsafe or unreliable for normalizing surface electromyography (sEMG) signals. To address this limitation, a normalized correction coefficient was introduced for pathological sEMG preprocessing, and an improved Hill-type muscle model (iHMM) was established to account for submaximal activation conditions. By combining inverse dynamics, static optimization, and a subject-specific lower-limb dynamic model, the proposed framework was used to estimate musculotendon force, knee joint torque, knee joint kinematics, and shank center-of-mass trajectory. In a preliminary validation involving six hemiplegic subjects, the predicted knee joint torques showed moderate to good agreement with the reference results, with correlation coefficients ranging from 0.724 to 0.807 and RMSE values ranging from 3.872 to 7.814 Nm. These preliminary results support the feasibility of the proposed framework for subject-specific biomechanical analysis of the hemiplegic lower extremity and suggest its potential utility in individualized rehabilitation assessment.