Technologies

21 pages, 1242 KB

Open AccessArticle

Mechanical and Tribological Properties of FFF-Printed TPU 95A with Uniform and Step-Gradient Infill Architectures

by Andrey P. Vasilev, Igor S. Makarov and Aitalina A. Okhlopkova

Technologies 2026, 14(9), 536; https://doi.org/10.3390/technologies14090536 (registering DOI) - 29 Aug 2026

This study evaluated the effects of uniform and symmetric step-gradient infill architectures on the density, mechanical properties, Shore D hardness, and dry-sliding coefficient of friction of FFF-printed thermoplastic polyurethane (TPU) 95A. Uniform triangular infills at 50%, 70%, and 90% were compared with a [...] Read more.

This study evaluated the effects of uniform and symmetric step-gradient infill architectures on the density, mechanical properties, Shore D hardness, and dry-sliding coefficient of friction of FFF-printed thermoplastic polyurethane (TPU) 95A. Uniform triangular infills at 50%, 70%, and 90% were compared with a surface-dense architecture (GI-1: 90–70–50–70–90%) and a core-dense architecture (GI-2: 50–70–90–70–50%). The experimental density of GI-2 (0.988 g/cm³) did not differ significantly from that of the uniform 70% configuration (0.990 g/cm³). GI-2 nevertheless exhibited significantly higher tensile strength (20.3 vs. 18.4 MPa) and a higher work of deformation to failure (48.3 vs. 32.3 MJ/m³). GI-1 had a lower experimental density than the uniform 90% configuration yet higher tensile strength and work of deformation to failure; the two configurations were not density-matched. Both gradient architectures exhibited lower mean compressive stresses at 20% strain than the uniform 70% and 90% specimens. GI-1 displayed a Shore D hardness comparable to the uniform 90% configuration and the lowest mean coefficient of friction (0.50 ± 0.01). Overall, step-gradient infill can improve specific tensile properties while allowing compressive compliance and the contact response of the outer shell to be tailored separately. However, the effects of gradient order and material content could not be fully separated because GI-1 and GI-2 differed in nominal mean infill density. Full article

(This article belongs to the Section Innovations in Materials Science and Materials Processing)

27 pages, 6197 KB

Open AccessArticle

Edge-Based Facial Emotion Recognition for Nurse-Assistive Robots Using a Compact CNN

by Quoc-Cuong Pham, Thanh-Long Le, Huy-Hoang Pham and Huu-Dung Nguyen

Technologies 2026, 14(9), 535; https://doi.org/10.3390/technologies14090535 (registering DOI) - 29 Aug 2026

Abstract

Facial emotion recognition (FER) can provide supplementary affective information for human–robot interaction, but deployment on resource-constrained assistive robots requires a balance between recognition performance and computational efficiency. This study presents an edge-based FER framework using a compact CNN operating on 48 × 48 [...] Read more.

Facial emotion recognition (FER) can provide supplementary affective information for human–robot interaction, but deployment on resource-constrained assistive robots requires a balance between recognition performance and computational efficiency. This study presents an edge-based FER framework using a compact CNN operating on 48 × 48 grayscale facial images and retaining all seven FER-2013 expression categories. Square-root-smoothed inverse-frequency weighting is employed to mitigate class imbalance without excessively emphasizing rare classes. On the held-out FER-2013 test set, the proposed model achieves 63.78% Accuracy and 59.32% Macro-F1, achieving higher Accuracy and Macro-F1 than the evaluated ImageNet-pretrained MobileNetV2 and MobileNetV3-Small baselines. INT8 post-training quantization reduces model size by 74.13% relative to FP32, with decreases of only 0.91 and 0.50 percentage points in Accuracy and Macro-F1, respectively. On a Raspberry Pi 3 Model B+, INT8 achieves a mean model-only inference latency of 19.30 ms and a model-only throughput of 51.82 FPS. Using an actor-disjoint RAVDESS protocol comprising 416 videos, EMA stabilization reduces prediction switching by 54.86% on the held-out test actors. These results support the feasibility of compact edge-based FER for assistive robotic interaction while emphasizing that the framework provides supplementary affective cues rather than clinical diagnosis or autonomous decision-making. Full article

(This article belongs to the Special Issue Advances in Automatics, Robotics & Artificial Intelligence)

► Show Figures

Figure 1

21 pages, 4251 KB

Open AccessArticle

Vis/NIR Spectral Sensing-Based Quality Prediction for Postharvest Sweet Potatoes

by Maoyuan Yin, Ruihua Zhang, Tianyu Zhu, Tao Sun, Wei Liu and Xinqing Xiao

Technologies 2026, 14(9), 534; https://doi.org/10.3390/technologies14090534 (registering DOI) - 29 Aug 2026

Abstract

Rapid and non-destructive assessment of sweet potato quality is important for postharvest management, processing suitability evaluation, and market quality control. In this study, a 12-channel visible/near-infrared (Vis/NIR) spectral sensing system was applied to predict multiple physicochemical quality attributes of postharvest sweet potatoes. Sixty [...] Read more.

Rapid and non-destructive assessment of sweet potato quality is important for postharvest management, processing suitability evaluation, and market quality control. In this study, a 12-channel visible/near-infrared (Vis/NIR) spectral sensing system was applied to predict multiple physicochemical quality attributes of postharvest sweet potatoes. Sixty independent sweet potato storage roots were measured at three representative positions, producing 180 position-specific observations; measurements from the same root were retained within the same validation group. The measured attributes included dry matter content (DMC), starch content (SC), soluble solids content (SSC), and the CIE 1976 L*a*b* (CIELAB) color coordinates L*, a*, and b*. Four spectral treatment conditions, including original spectra, normalization, standardization, and first-derivative transformation, were combined with partial least squares regression (PLSR), multiple linear regression (MLR), extreme gradient boosting (XGBoost), and random forest (RF), generating 16 prediction strategies for each quality attribute. Root-grouped five-fold cross-validation showed that the optimal models achieved coefficients of determination for cross-validation (R²CV) ranging from 0.9083 to 0.9190 and residual predictive deviation (RPD) values ranging from 3.3112 to 3.5230. Repeated grouped cross-validation produced mean R²CV values of 0.9113–0.9176, and root-block Y-scrambling yielded empirical p values of 0.005 for all six attributes. PLSR provided the highest cross-validated performance for all six quality attributes, although MLR showed comparable performance for several targets. These results provide preliminary evidence that discrete Vis/NIR spectral sensing can support simultaneous non-destructive estimation of multiple sweet potato quality attributes. External multi-batch and multi-cultivar validation is required before the models can be considered robust for practical deployment. Full article

(This article belongs to the Section Manufacturing Technology)

21 pages, 4427 KB

Open AccessArticle

Visible-Light-Driven Photocatalytic Degradation of Naproxen in Water by BiOCl_xI_1−x Solid Solutions: Performance, Operational Factors, and Mechanism

by Kun Fu, Huiping Deng, Pujing Yao, Pengkang Jin, Yuan Liu, Ning Luo and Huan Ma

Technologies 2026, 14(9), 533; https://doi.org/10.3390/technologies14090533 (registering DOI) - 28 Aug 2026

Abstract

The continuous release of pharmaceutical contaminants such as naproxen (NPX) into aquatic environments poses substantial ecological risks. In this study, a series of visible-light-responsive bismuth oxychloride-iodide (BiOCl_xI_1−x) solid solutions were synthesized via a simple one-step solvothermal method. XRD analysis [...] Read more.

The continuous release of pharmaceutical contaminants such as naproxen (NPX) into aquatic environments poses substantial ecological risks. In this study, a series of visible-light-responsive bismuth oxychloride-iodide (BiOCl_xI_1−x) solid solutions were synthesized via a simple one-step solvothermal method. XRD analysis confirmed the formation of a tetragonal matlockite-type solid solution, while SEM and TEM observations revealed three-dimensional flower-like hierarchical microspheres assembled from ultrathin nanosheets. Among the prepared samples, BiOCl_0.3I_0.7 exhibited the highest visible-light photocatalytic activity toward NPX degradation, achieving a removal efficiency of 87% within 60 min. Its apparent pseudo-first-order rate constant was 0.0740 min⁻¹, the highest among the investigated compositions. Experimental measurements showed composition-dependent band-gap narrowing, while representative DFT calculations indicated that I-for-Cl substitution modifies the valence-band electronic states, providing a qualitative electronic-structure explanation for the enhanced visible-light response. Evaluation of operational parameters showed that NPX degradation was favored at lower initial NPX concentrations and under acidic conditions, whereas humic acid and bicarbonate (HCO₃⁻) inhibited the process. TOC analysis further confirmed partial mineralization of NPX during photocatalysis. Electron paramagnetic resonance (EPR) analysis and reactive-species trapping experiments indicated that photogenerated holes (h⁺), singlet oxygen (¹O₂), and superoxide radicals (O₂^•−) were the dominant reactive species involved in NPX degradation. These findings demonstrate the potential of band-gap-engineered bismuth-based solid solutions for environmental remediation. Full article

(This article belongs to the Section Environmental Technology)

36 pages, 1630 KB

Open AccessArticle

A Techno-Economic Analysis of a Direct Vapour Generation Solar Cascade Organic Rankine Cycle for Efficient Cogeneration

by Xiao Ren, Zhaodong Tuo, Jing Li, Zhiying Zhang, Xiaolei Mou and Liang Gong

Technologies 2026, 14(9), 532; https://doi.org/10.3390/technologies14090532 (registering DOI) - 28 Aug 2026

Abstract

Concentrated solar power systems can provide dispatchable renewable energy, but their application in distributed cogeneration is constrained by high costs and thermal losses associated with indirect heat transfer. This study proposes a direct vapour generation cascade organic Rankine cycle (DVG-CORC) for combined heat [...] Read more.

Concentrated solar power systems can provide dispatchable renewable energy, but their application in distributed cogeneration is constrained by high costs and thermal losses associated with indirect heat transfer. This study proposes a direct vapour generation cascade organic Rankine cycle (DVG-CORC) for combined heat and power production. A biphenyl–diphenyl oxide (BDO) mixture is used as both the solar-collection fluid and the high-temperature cycle working fluid, while thermal storage and four operating modes are incorporated to accommodate variations in solar irradiance and enable continuous operation. Thermodynamic and economic models are developed to evaluate system performance under different evaporation and condensation temperatures. At an evaporation temperature of 400 °C, the maximum thermal efficiencies are 37.67%, 35.38%, and 33.11% at condensation temperatures of 60 °C, 80 °C, and 100 °C, respectively. At a condensation temperature of 60 °C, the cogeneration system generates an estimated annual revenue of USD 929,637, which is USD 284,269 higher than that of the power-generation-only configuration operating at a condensation temperature of 30 °C. These results demonstrate that direct vapour generation, cascade energy utilization, and heat recovery can improve the thermodynamic and economic performance of distributed solar cogeneration systems. Full article

(This article belongs to the Special Issue Solar Thermal Power Generation Technology)

29 pages, 8773 KB

Open AccessArticle

Sequence-Aware Dataset Auditing for Leakage-Free Benchmarking of YOLO Detectors for Bottle Detection

by Rafael Reveles-Martínez, Sebastián Burciaga-Sosa, José M. Celaya-Padilla, Salvador Castro-Tapia, Huizilopoztli Luna-García, Humberto Morales-Magallanes, Mayra N. Regalado-Pérez, César Landeros-Soriano, Umanel A. Hernández-González, Flabio D. Mirelez-Delgado and Hamurabi Gamboa-Rosales

Technologies 2026, 14(9), 531; https://doi.org/10.3390/technologies14090531 (registering DOI) - 28 Aug 2026

Abstract

This work presents a reproducible YOLO-based pipeline for bottle detection in sandy environments, emphasizing dataset integrity, leakage-free evaluation, and deployment-oriented model selection. A one-class dataset of 1585 images and 3167 annotated bottles was audited to identify annotation-format defects and near-duplicate contamination between training [...] Read more.

This work presents a reproducible YOLO-based pipeline for bottle detection in sandy environments, emphasizing dataset integrity, leakage-free evaluation, and deployment-oriented model selection. A one-class dataset of 1585 images and 3167 annotated bottles was audited to identify annotation-format defects and near-duplicate contamination between training and validation partitions. Sequence membership was reconstructed through perceptual-image similarity and used to assign complete image components to a sequence-aware train/validation split, eliminating the near-duplicate pairs found in the initial random partition. A controlled ablation holding model, seed, and corrected labels fixed showed that the random split reports 0.040 higher mAP@0.5:0.95 than the sequence-aware split (0.787 vs. 0.747), quantifying the leakage risk directly rather than only asserting it. Five YOLO configurations were then benchmarked under three independent seeds each; the observed mAP@0.5:0.95 differences among models (0.004–0.008) were small in absolute magnitude and, given only three seeds per model, are interpreted descriptively rather than as evidence of statistical equivalence or significance, so yolo11n_bottle was selected through a joint accuracy-parity, compactness, and exportability criterion (precision 0.982, recall 0.985, mAP@0.5 0.992, mAP@0.5:0.95 0.748), using approximately ten times fewer parameters than the largest configuration and producing a 5.2 MB checkpoint. ONNX export preserved detection geometry closely (100% count agreement, mean matched IoU

\geq 0.9998

), without meeting strict metric-parity tolerances. A stratified sample of 108 frames from operational RealSense BAG footage was manually annotated by an independent reviewer and evaluated quantitatively: mAP@0.5 remained close to the internal validation figure (0.927 vs. 0.992), while mAP@0.5:0.95 fell substantially (0.483 vs. 0.747), revealing a localization gap between the curated benchmark and operational conditions that this manuscript reports transparently. Together, these results show that dataset auditing, sequence-aware partitioning, multiseed benchmarking, and manually annotated operational evidence are each necessary to interpret a detection benchmark built from continuous video acquisition, providing a traceable, reproducible workflow for selecting and evaluating compact visual-perception models for resource-constrained environmental applications. Full article

(This article belongs to the Section Environmental Technology)

19 pages, 3823 KB

Open AccessArticle

A Multi-Generational YOLO Ensemble with Weighted Boxes Fusion for Robust Rescue-Oriented Object Detection in Chaotic Disaster Scenes

by Ming-Hseng Tseng and Yi-Wei Huang

Technologies 2026, 14(9), 530; https://doi.org/10.3390/technologies14090530 - 28 Aug 2026

Abstract

Accurate and robust object detection in complex disaster scenes is essential for effective emergency response; however, severe occlusion, dense overlap, and cluttered backgrounds pose significant challenges to conventional single-model detectors. To address these limitations, this study proposes a novel rescue-oriented detection framework that [...] Read more.

Accurate and robust object detection in complex disaster scenes is essential for effective emergency response; however, severe occlusion, dense overlap, and cluttered backgrounds pose significant challenges to conventional single-model detectors. To address these limitations, this study proposes a novel rescue-oriented detection framework that integrates a fine-grained disaster dataset, a cross-generational YOLO ensemble, and a consensus-based fusion strategy using Weighted Boxes Fusion (WBF). A dataset of 2323 images was constructed by re-annotating CDNIC19k with instance-level labels for four rescue-critical roles, enabling more precise evaluation in real-world scenarios. Heterogeneous YOLO models spanning multiple architectural generations were jointly exploited within a unified ensemble framework to leverage complementary representations. Meanwhile, a consensus-driven fusion strategy based on WBF was adopted to improve prediction aggregation in dense and occluded scenes. Experimental results showed that the proposed method outperformed single-model baselines and NMS-based approaches, improving mAP@0.5 from 0.696 to 0.756 (+6.0%) while maintaining strong recall and robustness. Analysis of the YOLOv12 family reveals an accuracy–efficiency trade-off, where lightweight models enable real-time inference while high-capacity models provide more reliable detection. Overall, these findings demonstrate that cross-generational architectural diversity combined with consensus-based fusion constitutes a generalizable and effective paradigm for high-precision disaster scene understanding under diverse deployment constraints. Full article

(This article belongs to the Special Issue Advanced Technologies in Computer Vision and Applications)

► Show Figures

Figure 1

19 pages, 3080 KB

Open AccessArticle

Data Symmetry Enhancement-Based Abnormal State Detection of High-End Hydrogen Compressors Under No-Fault Samples

by Fudong Li, Yue Shu, Bo Tao and Tianci Zhang

Technologies 2026, 14(9), 529; https://doi.org/10.3390/technologies14090529 - 27 Aug 2026

Abstract

Diaphragm-type hydrogen compressors serve as core equipment in hydrogen refueling stations, yet their early-stage deployment faces critical challenges, including insufficient fault mode data accumulation and unclear health evaluation criteria. Conventional manual operation and maintenance monitoring methods suffer from low efficiency and poor reliability, [...] Read more.

Diaphragm-type hydrogen compressors serve as core equipment in hydrogen refueling stations, yet their early-stage deployment faces critical challenges, including insufficient fault mode data accumulation and unclear health evaluation criteria. Conventional manual operation and maintenance monitoring methods suffer from low efficiency and poor reliability, failing to meet the demand for safe and stable operation. To address these limitations, this study proposes a novel abnormal state detection methodology applicable under no-fault sample conditions. The approach leverages data symmetry enhancement techniques to expand the training dataset using only normal operation records, and integrates multi-source sensor data for comprehensive equipment health analysis. Experimental results show that the proposed method achieves 96% detection accuracy, an approximately 8% missed detection rate, and millisecond computational latency, significantly outperforming traditional detection algorithms. This work provides a practical solution for early-stage equipment monitoring without fault samples, enhancing both technical robustness and operational efficiency for diaphragm hydrogen compressor maintenance. Full article

(This article belongs to the Topic Intelligent Maintenance and Health Management in Smart Manufacturing)

► Show Figures

Figure 1

29 pages, 2689 KB

Open AccessArticle

BAND: A Probabilistic Framework for Modeling Non-Stationary Heart Rate Variability in Rest–Stress–Rest Dynamics

by Matías Castillo-Aguilar, David Medina-Ortiz, Ruby Méndez Muñoz, Diego Mabe-Castro, Noah Beelders, Atenea Uribe-Ojeda, Marcelo A. Navarrete and Cristian Núñez-Espinosa

Technologies 2026, 14(9), 528; https://doi.org/10.3390/technologies14090528 - 27 Aug 2026

Abstract

Heart rate variability (HRV) forms the basis of non-invasive autonomic nervous system assessment. However, its analysis is constrained by the non-stationary nature of physiological signals. Standard analytical methods, which assume stationarity within fixed time windows, fail to capture dynamical effects of interest, such [...] Read more.

Heart rate variability (HRV) forms the basis of non-invasive autonomic nervous system assessment. However, its analysis is constrained by the non-stationary nature of physiological signals. Standard analytical methods, which assume stationarity within fixed time windows, fail to capture dynamical effects of interest, such as the response to a physiological stressor. This limitation obstructs the development of mechanistic hypotheses about autonomic control. Here, we address this challenge by introducing a probabilistic framework for modeling non-stationary HRV dynamics during transient, single-event perturbation-recovery paradigms. We propose a hypothesis-driven, generative model that transforms the physiological response into a continuous-time stochastic process controlled by a double-logistic function. This approach deconstructs the R-R interval (RRi) series into a set of interpretable parameters representing the latency, rate, and magnitude of distinct response and recovery phases. Through simulation, we show that the model achieves high-fidelity parameter recovery and describes these dynamics more accurately than conventional fixed-time window methods under conditions matching its own generative assumptions. We then apply the framework to an empirical exercise-recovery recording, generating a precise, falsifiable hypothesis of “dissonant autonomic recovery”, where the baseline RR interval and its variability recover to distinct extents. The biphasic autonomic non-stationary decomposition (BAND) framework provides a formal methodology for translating RRi time series into quantitative, testable estimates of their generative processes. Full article

(This article belongs to the Special Issue Assistive Technologies in Care and Rehabilitation: Research, Developments, and International Initiatives—Second Edition)

► Show Figures

Figure 1

23 pages, 5480 KB

Open AccessArticle

Prediction of Waterjet Cutting Depth Under Multi-Field Coupling Based on Zero-Shot Learning

by Feifei Lu, Yu Qiu, Dong Fan and Weiming Chen

Technologies 2026, 14(9), 527; https://doi.org/10.3390/technologies14090527 - 27 Aug 2026

Abstract

Sudden collapse accidents in mine roadways occur frequently, and post-disaster emergency rescue faces major challenges in terms of safety and efficiency. Therefore, efficient demolition equipment and intelligent prediction methods are urgently needed. Abrasive waterjet technology has considerable potential for complex disaster environments owing [...] Read more.

Sudden collapse accidents in mine roadways occur frequently, and post-disaster emergency rescue faces major challenges in terms of safety and efficiency. Therefore, efficient demolition equipment and intelligent prediction methods are urgently needed. Abrasive waterjet technology has considerable potential for complex disaster environments owing to its high efficiency, environmental friendliness, and cold-cutting characteristics. However, its cutting performance is affected by multiple coupled factors, including jet parameters, material properties, and environmental conditions. This makes accurate prediction difficult, especially under extreme or unseen operating conditions where available samples are limited. To address this problem, this study proposes a zero-shot learning-based multi-physics coupling prediction framework for the “jet–material–environment–effect” relationship. The framework is designed to predict abrasive waterjet cutting performance under unseen working conditions. First, a multi-factor cutting-performance dataset is constructed through a hierarchical experimental design. A generative adversarial network (GAN) is then introduced to expand the sample space and compensate for the discrete nature and limited distributional coverage of the experimental data. Second, a lightweight self-attention mechanism is employed to model high-dimensional input features globally, thereby improving the model’s ability to capture complex feature interactions. Finally, a joint loss function is designed to collaboratively optimize the generation and prediction processes. The experimental results show that the proposed model achieves a prediction accuracy of 98.3% on the test set, with a coefficient of determination R² of 0.967, outperforming WOA-SVM, BP neural network, EML, and Transformer models. The inference response time is approximately 3.2 s, indicating good engineering applicability. The results demonstrate that GAN effectively expands the sample space and improves model generalization, while the LightTransformer structure provides advantages in modeling high-dimensional coupled inputs. The proposed method can provide theoretical support and technical reference for intelligent demolition rescue and cutting-depth prediction under mine disaster conditions. Full article

(This article belongs to the Topic Responsible Classic/Quantum AI Technologies for Industrial Applications)

► Show Figures

Figure 1

36 pages, 12179 KB

Open AccessArticle

Evaluation of Drag Force for Selected .223 Rem Projectiles Across Different Flight Regimes

by Pavel Šafl, Jiří Maxa, Pavla Šabacká, Zdeněk Novotný, Robert Bayer, Tomáš Binar, Petr Bača, Jana Švecová, Jaroslav Talár and Robert Kutil

Technologies 2026, 14(9), 526; https://doi.org/10.3390/technologies14090526 - 25 Aug 2026

Abstract

This paper presents a study in the field of external ballistics focusing on the drag force of selected .223 rem projectiles. The drag force characteristics of projectiles with different geometric features were compared, including variations in projectile slenderness, ogive shape, boat-tail geometry, and [...] Read more.

This paper presents a study in the field of external ballistics focusing on the drag force of selected .223 rem projectiles. The drag force characteristics of projectiles with different geometric features were compared, including variations in projectile slenderness, ogive shape, boat-tail geometry, and projectile base design. The research combined theoretical analysis, experimental measurements, and CFD simulations. An initial assessment was performed to evaluate the influence of individual projectile geometry features on drag force across different flight regimes, namely the subsonic, transonic, and supersonic regions. These regimes exhibit fundamentally different aerodynamic characteristics, particularly due to the formation and evolution of shock waves, resulting in varying contributions of individual geometric features to the overall drag force. The results presented in this paper provide a foundation for future research. Subsequent studies will systematically investigate the influence of individual projectile geometry features on drag force in each flight regime, enabling a more comprehensive understanding of their aerodynamic significance. Full article

► Show Figures

Figure 1

38 pages, 25151 KB

Open AccessArticle

Prediction of Wing Pressure Distribution Using an Autoencoder-Based Surrogate Model

by Oleg Lukyanov, Damian Josue Guerra Guerra, Jose Gabriel Quijada Pioquinto, Nikolay Shevchenko, Evgenii Kurkin, Nguyen Hoang Le, Nikita Kuritsyn, Ivan Oseledets and Artem Nikonorov

Technologies 2026, 14(9), 525; https://doi.org/10.3390/technologies14090525 - 25 Aug 2026

Abstract

In the present work, an alternative methodology was developed for the rapid prediction of pressure distributions over wings of low-speed aircraft. A hybrid neural network architecture, named “MARTHA” (Model for Airloads Reconstruction using a Trained Hybrid Architecture), was presented, which is composed of [...] Read more.

In the present work, an alternative methodology was developed for the rapid prediction of pressure distributions over wings of low-speed aircraft. A hybrid neural network architecture, named “MARTHA” (Model for Airloads Reconstruction using a Trained Hybrid Architecture), was presented, which is composed of a Multilayer Perceptron and the decoder of an Autoencoder. Three compact representation models—Principal Component Analysis (PCA), Autoencoder (AE), and Variational Autoencoder (VAE)—were systematically evaluated to determine the optimal dimensionality reduction architecture; the AE was selected based on its superior reconstruction accuracy and training stability. The main feature of MARTHA is that it provides predictions of the differential pressure coefficient field in the form of monochrome images, where the pixel intensity directly represents the normalized pressure value. One of the main objectives of developing MARTHA was to create a rapid surrogate model that can approximate vortex lattice method (VLM) simulations in preliminary design and optimization tasks, particularly when thousands of wing configurations need to be evaluated. The key feature of the proposed model is its ability to predict the pressure distribution for trapezoidal wings of various geometries 10¹–10⁴ times faster than numerical models, while maintaining accuracy (R² = 0.9998). The data obtained are presented in a convenient format for their further use in CAE systems of strength analysis. To assess the practical utility of the proposed model, implementation cases were carried out using the finite element software ANSYS 18.2 for three wing configurations not present in the training dataset. The pressure fields predicted by MARTHA were mapped onto the wing meshes, and linear static structural analyses were performed. The obtained Von Mises stress distributions showed good agreement with the corresponding distributions obtained using numerical models. Full article

► Show Figures

Figure 1

27 pages, 2719 KB

Open AccessArticle

Driver Behavior Classification on Secondary Roads Using Machine Learning Models

by Albert Jose Potams, Raymond Ghandour, Zaher Al Barakeh and Karim Youssef

Technologies 2026, 14(9), 524; https://doi.org/10.3390/technologies14090524 - 25 Aug 2026

Abstract

Most existing driver behavior classification technologies have focused on highways and other primary road infrastructures, despite secondary roads accounting for a disproportionately large number of traffic fatalities worldwide. Compared with highways, secondary roads present greater variability in road geometry, infrastructure quality, and traffic [...] Read more.

Most existing driver behavior classification technologies have focused on highways and other primary road infrastructures, despite secondary roads accounting for a disproportionately large number of traffic fatalities worldwide. Compared with highways, secondary roads present greater variability in road geometry, infrastructure quality, and traffic interactions, making driver behavior recognition considerably more challenging. This paper investigates the classification of driver behavior on secondary roads using machine learning techniques. Naturalistic driving data obtained from the publicly available UAH-DriveSet dataset were analyzed using two complementary feature groups describing lane detection and traffic status. Four supervised machine learning algorithms, namely, Logistic Regression (LR), gradient boosting (GB), Random Forest (RF), and Artificial Neural Networks (ANNs), were evaluated to classify driving behavior into three categories: Normal, Aggressive, and Drowsy. The extracted features were first analyzed through statistical profiling and exploratory feature analysis before training and evaluating the classification models. The experimental results show that gradient boosting consistently achieved the highest performance for both feature groups, attaining an overall classification accuracy of approximately 67% while providing balanced precision, recall, and F1-scores across all behavioral classes. Logistic regression and random forest produced competitive but lower performance, whereas the Artificial Neural Network yielded the lowest classification accuracy. The obtained results demonstrate the effectiveness of ensemble learning methods for driver behavior recognition under secondary-road conditions and highlight their potential for integration into intelligent driver monitoring and Advanced Driver Assistance Systems (ADASs). By enabling earlier identification of aggressive and drowsy driving behaviors on secondary roads, the proposed approach could support timely driver warnings and safety interventions, potentially reducing accident risk. Furthermore, the findings provide a benchmark for future machine learning models designed for real-world secondary-road environments, where driving conditions are more variable and challenging than on highways. Full article

(This article belongs to the Special Issue Advanced Intelligent Driving Technology)

► Show Figures

Figure 1

25 pages, 3922 KB

Open AccessArticle

Neurocommunication and Affective-State Analysis: A Physiological-Proxy-Gated Bimodal Framework for Masked-Distress Scenarios

by Nayeli Bastidas-Benalcazar, Daniela Santos-Solís, Diego Cueva-Argudo, Enma Lechón-Santacruz, Diego Almeida-Galárraga, Paulo Navas-Boada, Henry Carvajal-Mora, Nathaly Orozco-Garzón and Andrés Tirado-Espín

Technologies 2026, 14(9), 523; https://doi.org/10.3390/technologies14090523 - 25 Aug 2026

Abstract

Facial displays can be ambiguous indicators of internal affect, but physiological-proxy-gated fusion has rarely been evaluated with transparent failure analysis. SilentDepress AI is examined here as an affect-fusion prototype rather than a clinical depression detector. We audited a local corpus of 30 video [...] Read more.

Facial displays can be ambiguous indicators of internal affect, but physiological-proxy-gated fusion has rarely been evaluated with transparent failure analysis. SilentDepress AI is examined here as an affect-fusion prototype rather than a clinical depression detector. We audited a local corpus of 30 video sequences from five participants, covering six nominal affective categories, and ran a publicly versioned seven-output facial classifier on every decoded frame. Mean frame logits produced a sequence-level accuracy of 20.0% (participant-block 95% CI 16.7–26.7%); the corresponding six-category restricted accuracy was 26.7%. We then performed 1000 paired constructed Monte Carlo cohorts per prespecified scenario using empirical visual Dirichlet priors and label-conditioned heuristic physiological proxies. In the primary sensitivity scenario, the highest mean constructed-case accuracy was 44.6% for 40/60 fusion. An independent external WESAD analysis using nested subject-level LOSO yielded 87.22% balanced accuracy (95% CI 77.22–95.56%) and 88.00% accuracy for native baseline-versus-stress discrimination. This external physiological evaluation was not synchronized with the local facial-video corpus and was not used as a seven-class fusion input. The simulation therefore evaluates the numerical behavior of the gating rule only and exposes the evidence still required for clinical evaluation. Full article

► Show Figures

Graphical abstract

29 pages, 4181 KB

Open AccessArticle

Open-Weight Multimodal LLMs Versus Manual Data Entry for Legacy ERP Digitization: A Comparative Evaluation of Accuracy, Cost, and Verifiability

by Chacharin Lertyosbordin and Boonyakorn Trangadisaikul

Technologies 2026, 14(9), 522; https://doi.org/10.3390/technologies14090522 - 24 Aug 2026

Abstract

Decades-old enterprise-resource-planning (ERP) systems lock operational data inside unstructured, human-readable reports, forcing slow, costly, error-prone manual re-keying. Because multimodal large language model (MLLM) capability is uneven, deploying MLLMs for extraction means trusting outputs without a labeled reference. We test this with a within-document [...] Read more.

Decades-old enterprise-resource-planning (ERP) systems lock operational data inside unstructured, human-readable reports, forcing slow, costly, error-prone manual re-keying. Because multimodal large language model (MLLM) capability is uneven, deploying MLLMs for extraction means trusting outputs without a labeled reference. We test this with a within-document controlled experiment on 400 controlled-substance stock-ledger documents (2951 records, 11 fields, predominantly Thai) from a Thai pharmaceutical factory, comparing trained human double-entry against four open-weight MLLMs (2 × 2 design: vendor × architecture) via OpenRouter. Human double-entry left 14 discrepancies against the adjudicated gold standard, none common to both operators. The strongest model, Qwen3-VL-32B-Instruct (Dense), reached 93.95% cell accuracy; among these four models, field accuracy varied more across vendors, whereas structural completeness differed consistently between dense models (0 missing records) and Mixture-of-Experts models (up to 51 of 2951 dropped). Deterministic accounting invariants flagged 0.61% of its records, leaving the unflagged majority 94.1% accurate across all 11 fields; adding calendar rules flagged 4.61% and raised residual date accuracy from 92.1% to 95.9%. We report both operating points and recommend the extended level where date fidelity is regulatory-critical. The pipeline is 13.5–29.4× faster in wall-clock terms and 97.5–99.5% cheaper. Gold-free, rule-based verification thus locates where MLLM reliability holds, giving human–AI collaboration quantified, disclosed residual risk rather than an implied guarantee. Even at the more conservative operating point, unflagged records average 94.5% accuracy across all 11 fields but only 43.7% on the free-text Remarks field, which the triage cannot check; the results support risk reduction and the localization of review effort, not unrestricted regulatory reliability across all fields. Full article

(This article belongs to the Special Issue Digital Data Processing Technologies: Trends and Innovations)

► Show Figures

Figure 1

22 pages, 4179 KB

Open AccessArticle

Posture-Constrained Workspace Analysis and Flow-Constrained Actuator-Space Time–Jerk Trajectory Planning for Heavy-Duty Hydraulic Demolition Robots

by Chentao Yao, Wendi Dong, Hui Zhang, Xingtao Zhang, Xizhong Cui, Zhuangwei Niu, Zheng-Yang Li, Jianwei Zhao, Dongjia Yan and Hongbo Li

Technologies 2026, 14(9), 521; https://doi.org/10.3390/technologies14090521 - 23 Aug 2026

Abstract

During high-speed multi-joint coordination, the nonlinear joint-to-cylinder mapping may increase the velocity and jerk peaks of the hydraulic cylinders, while simultaneous multi-cylinder motion may cause flow-peak superposition and increase the risk of exceeding the pump-flow limit. Addressing the limitations of traditional joint-space trajectory [...] Read more.

During high-speed multi-joint coordination, the nonlinear joint-to-cylinder mapping may increase the velocity and jerk peaks of the hydraulic cylinders, while simultaneous multi-cylinder motion may cause flow-peak superposition and increase the risk of exceeding the pump-flow limit. Addressing the limitations of traditional joint-space trajectory planning, which struggles to balance actuator-space smoothness, nonlinear inverse kinematics robustness, and multi-cylinder total-flow constraints, this paper proposes a multi-objective trajectory-planning method in the hydraulic-cylinder actuator space. First, a kinematic model is constructed based on the modified Denavit–Hartenberg method and hydraulic-cylinder closed-loop cosine mapping to evaluate effective moment arms and transmission sensitivity. Subsequently, a method combining Monte Carlo global search and Levenberg–Marquardt local iteration is adopted to solve inverse kinematics without explicitly computing the Moore–Penrose pseudoinverse of the Jacobian. On this basis, analytic quintic splines incorporating asymmetric perturbation terms are constructed, and a non-dominated sorting genetic algorithm II bi-objective optimization model for minimizing the motion time and the maximum absolute jerk in the actuator space is established, incorporating the total-flow hard constraint. Simulation results demonstrate that the motion time of the compromise solution is 7.71 s, the maximum absolute jerk in the actuator space is 22.94 mm/s³, and the total flow throughout the process is lower than 105 L/min. This method keeps the planned total-flow demand within the pump-flow capacity and reduces the risk that the planned actuator speeds cannot be maintained because of insufficient flow supply, providing a planning basis for the stable operation of heavy-duty hydraulic demolition robots. Full article

(This article belongs to the Special Issue Advances in Automatics, Robotics & Artificial Intelligence)

► Show Figures

Figure 1

29 pages, 12796 KB

Open AccessArticle

A Multi-Segment Fusion Architecture for Bone Age Estimation: Comparative Backbone Analysis and External Validation in a Single-Center Mexican Clinical Cohort

by Miguel A. Lozano-López, Daniel Román-Rojas, Jorge Gálvez and Aurora Espinoza-Valdez

Technologies 2026, 14(9), 520; https://doi.org/10.3390/technologies14090520 - 23 Aug 2026

Abstract

One of the most challenging tasks in pediatric medicine is bone age estimation from hand radiographs. Traditional bone age estimation approaches show limited performance across different demographic groups, making diagnostics prone to misclassification of growth abnormalities. On the other hand, convolutional neural networks [...] Read more.

One of the most challenging tasks in pediatric medicine is bone age estimation from hand radiographs. Traditional bone age estimation approaches show limited performance across different demographic groups, making diagnostics prone to misclassification of growth abnormalities. On the other hand, convolutional neural networks have demonstrated remarkable performance for many computer vision tasks, including bone age estimation. These models automatically extract and learn meaningful patterns capturing structural variations in bones. However, these models still present limited generalization capabilities, especially across different demographic populations, and are rarely evaluated beyond the public datasets on which they are trained. This paper presents a fusion architecture, F-DenseNet121, together with a systematic evaluation of its generalization across demographic populations. The employed methodology incorporates an anatomical segmentation strategy inspired by the Tanner–Whitehouse 3 (TW3) framework, combined with feature learning from the RSNA Pediatric Bone Age Challenge dataset. Instead of training a standalone convolutional model, the proposed methodology considers a convolutional network for each anatomical segment. This mechanism allows the model to learn localized skeletal patterns, reducing the influence of irrelevant structures. During training and validation with the RSNA dataset, the proposed F-DenseNet121 obtained a Mean Absolute Error (MAE) of 5.77 months during internal validation, a figure comparable to several reported convolutional models under their respective internal validation protocols. F-DenseNet121 was also evaluated using an independent 10.8% RSNA test validation subset, obtaining an MAE of 13.70 months, a result that remains substantially higher than published state-of-the-art benchmarks (4.2–6.2 months) and reveals a substantial generalization gap between internal validation and independent testing. To further examine this gap, external validation was performed using radiographs from Mexican patients. In this test, all convolutional models showed a significant performance difference between the public RSNA dataset and the clinical data from Mexican patients, with F-DenseNet121 and F-InceptionV3 achieving statistically indistinguishable external performance among the evaluated backbones. Rather than positioning these results as evidence of state-of-the-art accuracy, this study highlights that internal validation performance can substantially overestimate real-world reliability, and underscores the value of rigorous, multi-architecture comparison and external clinical validation for assessing the true applicability of automated bone age estimation systems. Full article

(This article belongs to the Special Issue Advanced Technologies of Biomedical Image Processing)

► Show Figures

Graphical abstract

8 pages, 187 KB

Open AccessOpinion

Robots in the OR: Hype, Hope, or Holding Pattern in Cholecystectomy?

by Muhannad Maher Abdin, Michael Connolly, Alden Stockam and Oleg Karaduta

Technologies 2026, 14(9), 519; https://doi.org/10.3390/technologies14090519 - 22 Aug 2026

Abstract

Robotic-assisted cholecystectomy (RAC) is increasingly presented as the next step in minimally invasive biliary surgery, but its added value over laparoscopic cholecystectomy (LC) remains uncertain. Comparative evidence suggests that RAC may reduce conversion to open surgery in some settings, yet it has not [...] Read more.

Robotic-assisted cholecystectomy (RAC) is increasingly presented as the next step in minimally invasive biliary surgery, but its added value over laparoscopic cholecystectomy (LC) remains uncertain. Comparative evidence suggests that RAC may reduce conversion to open surgery in some settings, yet it has not demonstrated consistent improvement in postoperative outcomes and is generally associated with longer operative time, higher costs, and unresolved safety concerns during dissemination and learning. RAC may offer value in selected complex cases and within structured robotic training programs, but these roles require subgroup-specific evaluation. For an established, safe, and efficient procedure such as LC, noninferiority is not enough; routine adoption should depend on demonstrable clinical, educational, or operational added value. Full article

(This article belongs to the Special Issue Technological Advances in Science, Medicine, and Engineering 2025)

27 pages, 1406 KB

Open AccessSystematic Review

Bridging the AI Language Divide: A Systematic Review of NMT and LLMs in Low-Resource Translation

by Sweeta Agrawal and Abayomi O. Agbeyangi

Technologies 2026, 14(8), 518; https://doi.org/10.3390/technologies14080518 - 21 Aug 2026

Abstract

The rapid evolution of AI-driven language technologies has inadvertently widened the gap between high-resource and marginalised languages. Despite significant progress in AI-driven translation for high-resource languages, low-resource languages remain underrepresented due to limited data, a lack of benchmarks, and evaluation challenges. This study [...] Read more.

The rapid evolution of AI-driven language technologies has inadvertently widened the gap between high-resource and marginalised languages. Despite significant progress in AI-driven translation for high-resource languages, low-resource languages remain underrepresented due to limited data, a lack of benchmarks, and evaluation challenges. This study presents a comprehensive systematic review of machine translation for low-resource languages, focusing on advances in neural machine translation (NMT) and large language models (LLMs) between 2017 and 2025. Following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, 63 studies were selected from the 1696 articles in the Scopus, Web of Science, and Google Scholar databases. The review identifies five dominant methodological approaches: data augmentation, back-translation, transfer learning, pre-training, and parameter-efficient fine-tuning. The findings reveal that model performance is highly dependent on resource availability: transformer-based NMT excels in moderate data settings, while LLMs demonstrate promising zero-shot and few-shot capabilities in extremely low-resource scenarios. Hybrid NMT–LLM approaches emerge as a particularly effective paradigm. The study also highlights critical challenges, including the absence of standardised benchmarks, over-reliance on inadequate evaluation metrics such as Bilingual Evaluation Understudy (BLEU), limited human evaluation, and significant geographic and linguistic underrepresentation. Additionally, ethical concerns related to bias, cultural representation, and community engagement are increasingly relevant. The findings contribute to advancing inclusive and equitable AI-driven language technologies. Full article

► Show Figures

Figure 1

Journal Description

Technologies

Latest Articles

Journal Menu

Journal Browser

Highly Accessed Articles

Latest Books

E-Mail Alert

News

Topics

Conferences

Special Issues

Topical Collections

Further Information

Guidelines

MDPI Initiatives

Follow MDPI