Search Results (241)

Abnormal gait associated with neuromuscular and musculoskeletal disorders represents a growing clinical burden, particularly in aging populations. This study presents a modular, low-cost Smart Rehabilitation Walker (SRW) that integrates multimodal sensing and real-time haptic feedback to enable simultaneous gait monitoring and corrective intervention in both clinical and home environments. The system combines force-sensing resistors for bilateral load symmetry assessment, inertial measurement units for fall detection, and surface electromyography (sEMG) for neuromuscular activity monitoring within a closed-loop assistive feedback architecture. A 15-day pilot study involving ten individuals with rheumatoid arthritis and clinically observed neurological gait abnormalities demonstrated measurable improvements in gait biomechanics. The Force Symmetry Index (FSI), calculated using the Robinson symmetry metric, decreased from an average of 0.9691 to 0.2019, corresponding to a 79.26% average reduction in inter-limb load asymmetry. Concurrently, sEMG measurements showed a substantial increase in neuromuscular activation (ΔEMG = 4.28), with statistical analysis confirming a significant improvement across participants (paired t-test: t(9) = 13.58, p < 0.001). To model rehabilitation trajectories, a nonlinear predictive framework based on Gaussian Process Regression achieved high predictive accuracy (R² ≈ 0.9, with a mean RMSE of 0.0385), while providing uncertainty-aware trend estimation. Validation using an independent amyotrophic lateral sclerosis gait dataset further demonstrated the transferability of the analytical pipeline. These results highlight the potential of sensor-enabled assistive walkers as scalable platforms for quantitative gait rehabilitation, adaptive feedback, and long-term mobility monitoring. Full article

Three-Dimensional (3D) face reconstruction from monocular Red-Green-Blue (RGB) imagery remains a fundamental yet ill-posed challenge in computer vision, with applications in biometrics, augmented reality/virtual reality (AR/VR), and intelligent visual sensing systems. While deep learning has significantly improved reconstruction fidelity and realism, existing surveys primarily focus on network architectures in isolation, often overlooking how sensing conditions, data acquisition protocols, and geometric calibration influence reconstruction reliability and evaluation outcomes. This paper presents a sensor-aware, end-to-end review of deep learning-based 3D face reconstruction and introduces a unified modular framework that connects sensing hardware, data acquisition, calibration, representation learning, and geometric refinement within a coherent pipeline. The reconstruction process is organized into four stages: sensor-driven acquisition and calibration, landmark estimation and feature extraction, 3D representation and parameter regression, and iterative refinement via differentiable rendering. Within this framework, we examine how sensor characteristics, calibration accuracy, representation models, and supervision strategies affect reconstruction accuracy, perceptual quality, robustness, and computational efficiency. We further synthesize the reported results across widely used benchmarks using both geometric and perceptual metrics, highlighting trade-offs between reconstruction fidelity and deployment constraints. By integrating sensing-aware analysis with architectural evaluation, this survey provides practical insights for developing scalable and reliable 3D face reconstruction systems under real-world conditions. Full article

In the age of remote communication and digital archiving, automated analysis of voice data has become increasingly important in various application areas. Despite significant advances in the field of Automatic Speech Recognition, integrating speaker recognition, textual sentiment analysis, and acoustic sentiment detection within a unified real-time processing pipeline remains a challenging task. Current approaches are often limited to monolithic designs or operate in batch processing modes, which restricts their scalability and real-time applicability. To address this gap, this work proposes a novel feature selection method called RAM, along with a hybrid decision-level merging approach combining Conv1D CNN and AutoML-based models. The proposed hybrid framework enables independent model training and integrates its probabilistic outputs through a weighted merging strategy for performance improvement. Furthermore, a scalable microservice-based software architecture has been developed to support real-time processing, feature selection, and model deployment. This design enhances system modularity, flexibility, and integration capability in practical applications. Experimental results show that when the proposed RAM method is used in conjunction with a hybrid AI model, it achieves over 97% accuracy in speaker recognition and over 82% accuracy in emotion classification, even with short audio samples. These findings demonstrate that the proposed approach provides a robust and efficient solution for real-time speech analysis tasks. Full article

Building upon previously developed thin-film modular soft actuators for elongation and deflection, this study develops a modular soft robot for pipeline locomotion, addressing insufficient anchoring capability in confined environments. Conventional inflatable airbags typically expand into spindle-shaped geometries, resulting in limited contact length and reduced effective gripping stability. To overcome this issue, a corrugated thin-film gripping actuator is proposed, in which two high-aspect-ratio sub-airbags are arranged above a compression structure to regulate deformation through geometric constraints. Numerical simulation and experimental evaluation were conducted to investigate contact behavior and locomotion performance. Under an input pressure of 30 kPa, the proposed design achieves a contact length of 46 mm, compared to 37 mm for a conventional three-layer airbag configuration under the same conditions, corresponding to a 24.33% increase in a 10 mm plate-spacing environment. The gripping module is integrated into the modular framework to extend the motion primitives of the soft robot to include anchoring functionality. The results indicate that the corrugated structure effectively suppresses the spindle effect and improves contact effectiveness under compression. These findings demonstrate that structural regulation of thin-film pneumatic actuators provides a feasible strategy for enhancing anchoring performance and locomotion capability of soft robots in confined pipeline environments. Full article

Smart-city IoT ecosystems depend on a large number of devices with limited resources, which often lack built-in security mechanisms. While traditional cloud-based or gateway-centric intrusion detection systems (IDSs) offer essential security, they are still characterized by high detection latency, considerable bandwidth demand, and a lack of precise monitoring of single device actions. This study proposes the Edge AI Bridge, a novel micro-computing security layer positioned between IoT devices and the gateway to enable early-stage threat interception. The architecture integrates embedded AI hardware with a hybrid pipeline, utilizing unsupervised anomaly detection for behavioral profiling and a lightweight signature-matching module to minimize false positives. System operations—including localized traffic inspection, protocol parsing, and feature extraction—are performed before data aggregation, which preserves device-level privacy and reduces the computational burden on the IoT gateway. The contemporary CIC-IoT-2023 dataset, which captures a wide range of smart-city protocols and attack vectors, is used to evaluate the architecture. The Edge AI Bridge leads to a significant reduction in detection latency—≈50 ms on average as opposed to the 500 ms of cloud-based solutions—while the resource footprint is kept low to about 20% CPU utilization. The Edge AI Bridge demonstrates a potential solution that is scalable, modular, and can preserve privacy while improving the cyber resilience of the smart-city infrastructures that are large, heterogeneous, and difficult to manage. Full article

This paper presents a hardware-accelerated signal processing architecture for OFDM-based vehicular networks that integrates crypto-agile adaptive encryption on a Xilinx Kintex-7 FPGA. The encryption layer is tightly coupled to the OFDM modulation/demodulation pipeline, enabling secure real-time signal processing for V2X communications without disrupting the baseband chain. A context-aware pre-selection unit dynamically selects among hardware cipher primitives based on latency constraints, security requirements, and channel conditions. The current prototype implements and synthesizes AES-128 as the primary block cipher, while ASCON (NIST lightweight AEAD) and Keccak (SHA-3 foundation) are validated through RTL simulation and architectural integration, demonstrating crypto-agility across block, AEAD, and sponge-based primitives. DES is retained solely as a legacy reference for backward-compatibility evaluation and is not recommended for secure V2X deployment. The design adopts a modular decoupling strategy in which cryptographic engines interface with a unified buffering and interleaving subsystem, enabling hardware-based single-cycle cipher switching without partial reconfiguration. FPGA results demonstrate sub-microsecond cryptographic processing latencies with moderate resource utilization, preserving the timing budget of latency-sensitive vehicular services. AES-128 provides standard-strength encryption, while ASCON and Keccak offer lightweight and sponge-based alternatives suited to constrained IoV platforms. Specifically, the implemented AES-128 core achieves a throughput of 1.02 Gbps with a switching latency of 86 ns, verified across 10 randomized transitions with a 99.99% success rate and zero data corruption. The ASCON and Keccak cores attain throughput-to-area efficiencies of 2.01 and 1.47 Mbps/LUT, respectively, at a unified clock frequency of 50 MHz. All acronyms are defined at first use and a complete list of abbreviations is provided prior to the reference section. Full article

Airborne particulate matter (PM) poses a major health risk, yet species selection for vertical greening systems (VGS) is poorly quantified. We evaluated PM retention by seven commercially available shade-tolerant species grown in a modular living wall system (LWS) on a north-facing façade at Sapienza University of Rome. After 3 months of in situ exposure, leaves were analyzed via SEM (1000×), collecting 210 images, 30 per species. An automated FIJI/ImageJ pipeline segmented particles, computed equivalent circular diameters, and classified them into (PM < 0.5, PM [0.5, 1), PM [1, 2.5), PM [2.5, 10), and PM ≥ 10 µm). Across species, ultrafine and fine fractions dominated deposits, with the <0.5 µm class typically comprising 60–70% of counts. Vinca minor cv. albomarginata exhibited the highest densities in ultrafine and fine classes, closely followed by Fatsia japonica; Hedera helix captured more coarse particles (2.5–10 µm and >10 µm). Heuchera sanguinea consistently displayed the lowest densities across all size classes. Performance patterns aligned with leaf surface traits: wax-coated, moderately rough or gently structured cuticles favored adhesion, whereas highly irregular microrelief did not consistently enhance retention. Methodological considerations include thresholding sensitivity, use of equivalent circular diameter for irregular particles, and an upper area filter that may undercount large aggregates. The findings identify Vinca minor cv. albomarginata and Fatsia japonica as priority species for PM mitigation in shaded VGS, with Hedera helix complementing coarse PM capture. The results provide trait-based, design-oriented guidance for living wall species selection in Mediterranean urban and indoor contexts. Full article

Handball performance analysis is still often conducted through the manual review of match videos, while automation on broadcast footage remains challenging due to camera motion, strong perspective effects, and frequent occlusions during dense interactions. This study presents a practical and reproducible monocular pipeline for extracting handball analytics from a single broadcast viewpoint. Players are detected per frame, tracked over time, and projected onto a standardized handball court via homography-based camera calibration. The resulting court-referenced trajectories in metric units enable motion indicators such as distance covered and speed, along with coaching-oriented visual summaries, including trajectory overlays and heatmaps. In addition, clip-level action recognition is performed using interpretable kinematic and scene-derived features and lightweight classifiers, with a comparative evaluation across multiple classical models. The modular design keeps the intermediate steps explicit, supports reproducibility, and facilitates interpretation of both intermediate outputs and final analytics. Experiments on the UNIRI handball dataset demonstrate that meaningful performance analytics and action understanding can be obtained from single-camera broadcast video using transparent intermediate representations. This work highlights the practical potential of interpretable trajectory-based modeling for under-instrumented sports and provides a reproducible baseline for future extensions incorporating richer contextual cues. Full article

Flexible automated assembly lines (FAALs) in Industry 4.0 require robust quality management that integrates operational data with systematic risk analysis. However, Process Failure Mode and Effects Analysis (PFMEA) documents are often developed during the design phase and not systematically updated with actual production data, leading to a gap between formal risk assessment and operational reality. This study addresses this gap by developing and validating an integrated data-driven framework that combines classical quality tools (process flow charts, check sheets, cause-and-effect diagrams, and Pareto analysis) with data-driven PFMEA, creating traceable links from operational logs to risk ratings. While individual quality tools are well-established, the core contribution of this work is a structured data transformation pipeline that creates traceable, auditable linkages from raw operational event logs to calibrated PFMEA ratings with quantified uncertainty—a combination not previously demonstrated for flexible assembly systems. The framework was applied to FMS-200, a modular FAAL for bearing units, consisting of eight stations and a common transfer system. Analysis of 186 failure events across 2743 assembly cycles, including 18 product configurations, identified 40 distinct failure modes with risk priority number (RPN) values ranging from 60 to 378, revealing that approximately 90% of the aggregated risk is associated with pneumatic systems. Monte Carlo uncertainty analysis (10,000 iterations) demonstrated robust rank stability, with the top five failure modes maintaining their relative ordering in over 90% of simulations. The framework provides production and quality managers with a systematic methodology to maintain PFMEA relevance through continuous data integration, enabling evidence-based prioritization of improvement actions. Full article

Background: A major technical challenge in single-cell transcriptomics is the absence of an integrative analytic pipeline that can simultaneously leverage gene regulatory network (GRN) architecture, AI-assisted gene panel discovery, and functional relevance analyses to generate coherent biological insights. Existing approaches often treat these components independently, focusing on clusters, marker genes, or predictive features without integrating them into a mechanistically grounded framework. Consequently, comprehensive screening that links regulatory association, gene signature screening, and functional interpretation within single-cell datasets remains limited, underscoring the need for an integrated strategy. Methods: We developed an integrative bioinformatics pipeline based on Gene regulatory network–AI–Functional Analysis (GAFA), combining latent-space integration, unsupervised clustering, diffusion pseudotime analysis, lineage-resolved generalized additive modeling, GRN inference, and machine learning-based gene panel discovery. This framework enables systematic mapping of cell-state structure, reconstruction of differentiation and effector trajectories, and identification of transcriptional and regulatory features strongly associated with clinical outcomes. As a case study, we applied the pipeline to NK cell transcriptomes from six CML patients (two early relapse, two late relapse, two durable treatment-free remission—TFR; 15 samples) collected at TKI discontinuation and 6–12 months after therapy cessation. Results: We reanalyzed publicly available scRNA-seq data from a previously published CML cohort to evaluate NK-cell transcriptional programs associated with treatment-free remission and relapse. We resolved six transcriptionally distinct NK cell states spanning CD56^bright-like cytokine-responsive, early activated, terminally mature, cytotoxic, lymphoid trafficking, and HLA-DR⁺ immunoregulatory populations, each exhibiting outcome-specific compositional differences. Pseudotime analysis revealed two major NK cell lineages—a maturation trajectory and a cytotoxic effector trajectory. TFR samples displayed balanced occupancy of both lineages, whereas early relapse samples showed marked depletion of the maturation branch and preferential accumulation in cytotoxic end states. AI-guided feature selection and random forest modeling identified an 18-gene panel that distinguished NK cells from TFR and relapse samples in an exploratory manner. Among them, CST7, FCER1G, GNLY, GZMA, and HLA-C were conventional NK-associated genes, whereas ACTB, CYBA, IFITM2, IFITM3, LYZ, MALAT1, MT2A, MYOM2, NFKBIA, PIM1, S100A8, S100B, and TSC22D3 were novel. The GRN inference further uncovered outcome-specific regulatory modules, with RUNX3, EOMES, ELK4, and REL regulons enriched in TFR, whereas FOSL2 and MAF regulons were enriched in relapse, and their downstream targets linked to IFN-γ signaling, metabolic reprogramming, and immunoregulatory feedback circuits. Conclusions: This AI-enabled single-cell analysis demonstrates how NK cell state composition, differentiation trajectories, and regulatory network rewiring collectively shape TFR versus relapse following TKI discontinuation in CML. The integrative pipeline provides a modular framework that could be extended to additional datasets for data-driven biomarker discovery and mechanistic stratification, and highlights candidate transcriptional regulators and NK cell programs that may be leveraged to improve remission durability, pending validation in larger patient cohorts. Full article

Effective wildfire management demands seamless integration of real-time detection and long-term spread forecasting. This paper proposes a novel power-efficient UAV–satellite hybrid pipeline that synergizes the agility of UAVs with the scale of satellite intelligence. The system begins with a dashboard-guided, multi-UAV detection module that scores fire likelihood from historical satellite data and enables scalable, energy-efficient deployment with low-latency onboard processing. This aerial component ensures persistent surveillance and reliable ignition detection, supported by a Dual LoRa (Long Range) communication scheme for robust and low-power connectivity. It achieves an F1-score of 97.4% while minimizing power consumption to extend operational flight times. Following detection, the pipeline transitions to a dynamic perimeter-prediction phase utilizing a custom Canadian boreal dataset. We employ a Squeeze-and-Excitation Residual U-Net (SE-ResUNet) to model spatiotemporal fire propagation based on static terrain and dynamic environmental features. The model was validated using a dynamic simulation framework that evaluates temporal consistency and convergence behavior against final cumulative burned-area masks, effectively addressing the absence of daily ground truth. Under these conditions, the model achieves a recall of 84% and an AUC of 0.97, demonstrating a strong capability to delineate active fire fronts. By coupling dashboard-driven UAV sensing with satellite-based predictive modeling, this work establishes a modular, foundational framework to support data-scarce forecasting in modern wildfire management. Full article

This paper presents a conceptual architecture for an AI-Driven Adaptive Encryption Device (AI-AED), a tri-modular hardware platform embodied in a registered industrial design. The device integrates a Secure Input Module, an AI-Enhanced Central Processing Unit with biometric authentication, and a Secure Output Module connected by unidirectional buses. We formalise the adaptive encryption policy as a constrained Markov decision process (CMDP) over a discrete action space of 216 cryptographic configurations, with safety constraints that provably prevent convergence to insecure states. A formal threat model based on extended Dolev–Yao assumptions with four physical access tiers defines attacker capabilities, and anti-downgrade safeguards enforce a monotonically non-decreasing security floor during threat escalation. An information-theoretic analysis shows that adaptive algorithm selection contributes an additional entropy term H(α) to ciphertext uncertainty, upper-bounded by log₂(|L_enc|) ≈ 1.58 bits, while noting this represents increased attacker uncertainty rather than a strengthening of any individual cipher. A component-level latency model estimates 0.91–1.00 ms pipeline latency under normal operation and 3.14–3.42 ms under active threat, including integration overhead. Simulation validation over 1000 episodes compares a tabular Q-learning baseline against the proposed Deep Q-Network operating on the continuous state space: the DQN achieves 82% fewer constraint violations, 6× faster threat response, and more stable policy switching, demonstrating the advantage of continuous-state reinforcement learning for safety-critical adaptive encryption. All claims are positioned as theoretical contributions requiring empirical validation through prototype implementation. Full article

Speech-to-text translation (S2TT) for low-resource languages remains challenging due to the scarcity of parallel speech translation data and the susceptibility of modular pipelines to error propagation. This paper presents a controlled empirical comparison of cascade and end-to-end approaches for Kazakh–Russian speech translation using the ST-kk-ru dataset (≈332 h, 140 k triplets). The cascade framework is strengthened with recent pre-trained models for automatic speech recognition and neural machine translation, achieving 21.3 BLEU on the test set. Three representative end-to-end architectures are evaluated under identical data conditions. The strongest direct model, combining a Wav2Vec 2.0 encoder with an mBART decoder augmented by a length adaptor and adapter modules, reaches 17.97 BLEU, compared with 15.35 BLEU for FAIRSEQ S2T and 16.3 BLEU for ESPnet-ST. Automatic evaluation is complemented by expert manual assessment and targeted linguistic analysis. Results indicate that, under current low-resource conditions, cascade systems provide higher translation accuracy and better morpho-syntactic fidelity, while end-to-end models remain competitive and offer advantages in architectural simplicity and potentially reduced inference latency (due to single-pass processing), although empirical measurements were not conducted in this study. This study establishes a reproducible benchmark for Kazakh–Russian speech translation and highlights practical trade-offs between modeling paradigms in low-resource, morphologically rich settings. Full article

The rapid growth of 5G necessitates wireless receivers capable of high-speed, low-latency communication under complex channel conditions. Traditional receivers struggle with the performance–complexity trade-off in massive MIMO systems, where linear detectors underperform and maximum likelihood (ML) detection becomes computationally prohibitive. Deep-learning-based model-driven approaches have demonstrated a favorable balance between detection performance and computational cost. However, despite their algorithmic promise, the transition of these learned detectors into practical, real-time systems is critically hampered by inefficient hardware mapping, resulting in suboptimal throughput, high resource overhead, and limited scalability. To bridge this gap, this paper presents LwAMP-Net, a dedicated FPGA accelerator for a lightweight learned AMP detector. We propose a modular and multi-mode hardware architecture for LwAMP-Net, featuring an outer-product-based dataflow that mitigates pipeline stalls and multi-mode processing elements that adapt to diverse computation patterns. These innovations jointly enhance computational parallelism and resource utilization on the FPGA. Implemented on a Xilinx XC7VX690T FPGA for a 128 × 8 MIMO system with 16QAM, the accelerator achieves a 49.2% higher normalized throughput per iteration, an 85.4% improvement in throughput per LUT slice, and a 12.7% improvement in throughput per DSP compared to the state-of-the-art methods. This work provides a complete architectural solution for deploying high-performance, hardware-efficient learned MIMO detectors in real-world systems. Full article

A stereo vision framework is designed to improve disparity estimation in occluded and boundary regions, targeting autonomous driving scenarios. The proposed architecture combines frozen Distilled Internally Normalized Optimized Version 2 features with a modular three-stage attention fusion strategy, which consists of bottom-up semantic propagation, top-down detail enhancement, and cross-view attention mechanisms. These stages jointly enforce semantic consistency, structural integrity, and accurate correspondence modeling. The fused features are then processed by an Iterative Geometry Encoding and Volumetric regression-based disparity estimation module for multi-stage regression and iterative refinement. A three-phase training pipeline is employed, including pretraining on SceneFlow, fine-tuning on virtual Karlsruhe Institute of Technology and Toyota Technological Institute (KITTI) benchmarks, and adaptation to the KITTI and ETH Zurich 3D benchmark dataset. The model achieves an out-of-center, non-occluded pixel error of 7.45% on KITTI2012 and a D1-all error of 4.10% on KITTI2015. Beyond quantitative performance, the proposed method produces visually superior disparity maps. The enhancements of boundary sharpness, occlusion completion, and structural coherence demonstrate the strong potential of the proposed algorithm for real-world deployment in dynamic and complex environments. Full article