Search Results (10,220)

Vision-and-language navigation (VLN) traditionally relies on explicit reasoning chains, which, despite being interpretable, impose severe constraints on inference efficiency and scalability in long-range environments. Existing multimodal large language models (MLLMs) frequently encounter latency bottlenecks due to the generation of verbose textual narratives during decision-making. To address these limitations, we propose spatial reasoning vision-and-language navigation (SR-VLN), a novel framework that shifts the paradigm from explicit chain-of-thought (CoT) to an implicit spatial representation space. SR-VLN introduces a pyramidal hierarchical history framework integrated with perceptual compression to condense historical trajectories into multi-scale representations, effectively minimizing token overhead while preserving critical spatial semantics. Rather than generating verbose textual reasoning steps, SR-VLN employs compact, learnable spatial tokens (S-Tokens) to perform agile inference directly within the latent feature space. To establish robust causal mappings between these implicit states and navigational actions, we employ a hybrid training strategy that combines sparse reward supervision with reinforcement learning via GRPO. Extensive evaluations on the R2R, REVERIE, and SOON datasets demonstrate that SR-VLN achieves state-of-the-art overall navigation performance, while maintaining a comparable balance between accuracy and efficiency. Compared to explicit reasoning baselines, our method reduces token consumption by 68% and achieves a 4.1× speedup in inference while reaching a 76.02% success rate and a 73.80% SPL on the R2R unseen split, thereby facilitating near-real-time action prediction in long-range navigation environments. Full article

Smartwatches are increasingly used in safety-critical scenarios, yet their optical heart-rate (HR) measurements often contain noise, artifacts, and missing data, undermining clinical trust. This paper presents Edge Beats, a data-curation layer and end-to-end architecture that enables the low-cost, open source PineTime smartwatch to function as a practical HR sensing node for distributed wearable systems. Heart-rate packets are streamed from PineTime to an ESP32 at the edge layer over Bluetooth Low Energy (BLE), then forwarded via an embedded Message Queuing Telemetry Transport (MQTT) broker to an edge server laptop for processing and visualization. A lightweight multi-stage algorithm cleans and smooths the HR stream using physiological boundary checks, a configurable data imputation technique, and exponential moving average (EMA) smoothing, all designed for real-time operation on resource-constrained hardware. We have evaluated the system over long monitoring sessions and compared the processed PineTime output against a commercial Huawei GT Pro 2 smartwatch. The system suppresses extreme spikes and short-term oscillations, yielding a more stable HR trace with qualitative agreement to the reference trends while keeping values in a physiologically plausible range. Network measurements show low latency (almost 3 ms one-way, 15 ms RTT) and stable throughput, and power measurements (100–450 mW for ESP32 and 3–70 mW for PineTime watch) confirm that continuous HR streaming over BLE and MQTT is feasible within the PineTime’s energy budget. These results imply that data stream processing combined with a modest publish–subscribe architecture improves the stability and usability of HR streams obtained from commodity wearable sensors, making PineTime a candidate as a complementary component for mission-critical health and safety systems. Full article

Background: Sepsis remains a leading cause of morbidity, mortality, and healthcare expenditure worldwide. Despite international guidelines and diagnostic criteria, real-world variability in coding, treatment, and outcomes persist. This retrospective study analyzed 13,994 coded sepsis-related hospitalizations identified through administrative ICD-9-CM algorithms between 2016 and 2024 to evaluate the burden of sepsis, temporal trends, clinical outcomes, and healthcare costs within a regional health system. Methods: Hospitalization data across four local health authorities (ASL 201–204) over an 8-year period were analyzed. The coded sepsis cases were identified using validated ICD-9-CM-based algorithms and classified into four groups according to available microbiological coding: Gram-positive, Gram-negative, anaerobic and unspecified. Variables included patient demographics, length of stay, costs, outcomes (in-hospital and post-discharge mortality) and presence of septic shock. Comparative analyses were conducted using descriptive statistical methods and One-way ANOVA test and chi-squared tests were applied to evaluate the significance of differences. Multivariable logistic regression models were used to identify independent predictors of 6- and 12-month mortality. Results: The dataset included 13,994 coded sepsis-related hospitalizations, with the largest subgroup being ‘unspecified’ (48.0%). Among cases with specified etiology, coded anaerobic sepsis categories, though rare (0.7%), were associated with higher in-hospital mortality (45.5%) and economic burden (avg. € 8563). Mortality remained high at 6 and 12 months across all types, exceeding 50% post-discharge. Increasing age (OR ≈ 1.06 per year) and septic shock (OR ≈ 4.5–4.8) were the strongest independent predictors of mortality. Differences across microbiological groups should be interpreted cautiously given the high proportion of cases without organism-specific coding. Despite a modest reduction in mortality over time, sepsis was associated with persistently high 6- and 12-month mortality, highlighting a substantial long-term burden beyond the acute phase of illness. These findings suggest that sepsis-related hospitalizations are associated with substantial long-term mortality beyond the acute phase of illness. Discussion: These findings underscore the clinical and economic impact of sepsis in hospitalized patients, across microbiological coding categories. The high mortality rate at 6–12 months may support the need for further investigation into structured post-discharge follow-up strategies. Sepsis represents a substantial clinical and economic burden within the regional healthcare system, with persistently elevated short- and mid-term mortality. Incomplete organism-level documentation limits direct etiologic comparisons and highlights the need for improved integration between clinical, microbiological, and administrative data systems. Future research should integrate clinical variables and lab results to enable risk stratification and intervention planning. Full article

Recognizing student behaviors in classroom videos remains challenging due to complex backgrounds, frequent occlusions, subtle inter-class motion differences, and temporal jitter in frame-wise predictions. To address these issues, this paper proposes a hybrid student behavior recognition framework that integrates a Multi-branch Spatiotemporal Attention Network (MSTA-Net) with a Behavior State Kalman Filter (BSKF). At the perceptual level, MSTA-Net employs decoupled channel, spatial, and short-term temporal attention branches to enhance discriminative behavioral features while suppressing irrelevant background information. At the cognitive level, BSKF reformulates behavior recognition as a continuous state estimation problem in a high-dimensional probability space, where behavioral inertia is exploited to smooth noisy observations and improve temporal consistency. Experimental results on the SCB-Dataset and real-world classroom video sequences demonstrate that the proposed method achieves an accuracy of 94.7% and a real-time inference speed of 33 FPS. Compared with purely deep learning-based models, the proposed framework reduces the Action Category Switching (ACS) rate by 50%, indicating substantially improved robustness in long-term behavior recognition. These results suggest that coupling attention-based perception with Kalman-based state estimation provides an effective and efficient solution for reliable student behavior analysis in intelligent classroom environments. Full article

Urban street lighting remains a significant source of energy consumption in cities, largely due to static operation and limited responsiveness to real-time conditions. This inefficiency increases operational costs and environmental impact, especially in rapidly urbanizing regions. To address this issue, this study investigates IoT-enabled smart street lighting as an adaptive and data-driven solution within smart city frameworks. The work focuses on the growing body of research in this domain and examines its evolution, technical structure, and emerging environmental role. The study aims to provide a structured synthesis that connects research trends with system-level design, while highlighting the transition from energy-focused systems to multifunctional urban platforms. A bibliometric-driven and thematic review approach is adopted. A dataset of 151 publications was analyzed using Bibliometrix and Biblioshiny tools to extract trends, collaboration patterns, and research themes. This analysis is complemented by a qualitative evaluation of system architectures, sensing technologies, communication models, and control strategies. The findings indicate a sustained annual growth rate of 14.87% and a highly collaborative research landscape, with an average of 3.97 authors per study. The results also reveal that energy efficiency remains the dominant focus, while environmental integration is emerging but still underrepresented. The study further identifies key gaps related to scalability, sensor reliability, and the lack of standardized evaluation metrics. The outcomes provide a comprehensive roadmap for future research and support the development of scalable, intelligent, and sustainable lighting systems. The proposed insights are applicable to urban environments globally, particularly in regions seeking cost-effective and energy-efficient infrastructure solutions. Full article

Precise dimensional conformity of precast concrete members is critical for structural performance and on-site assembly accuracy, yet conventional manual inspection remains labor-intensive and unable to scale with modern production-line throughput. Existing scan-vs-BIM approaches address geometric verification in principle but are constrained by manual registration dependencies, the absence of machine-readable IFC-linked tolerance criteria, and limited validation under real factory yard conditions. This study presents a production-scale automated shape quality control (SQC) framework that closes all three gaps simultaneously. A purpose-designed two-point target device enables fully automated, repeatable registration seed-point extraction. A formal IFC property-set-linked rule engine architecture—comprising entity extraction, deviation computation, rule interpretation, and pass/fail decision stages—replaces ad hoc script-based tolerance checking with an interoperable, auditable compliance pipeline. Factory-scale validation on precast arch segments (n = 10) and wall panels (n = 12) achieved registration RMSE of 1.25–1.95 mm, pass rates exceeding 91%, and a 37.1% reduction in inspection time versus manual methods (95% CI: 34.5–39.6%; p < 0.001; Cohen’s d = 3.89). Repeatability testing yielded ICC = 0.971 and Bland–Altman limits of agreement of [−0.45, +1.07] mm. The framework represents a substantive step toward fully digital, production-integrated quality management for industrialized precast construction. Full article

To address the joint requirements of illegal parking detection and license plate recognition under complex traffic scenarios and limited edge-device resources, this study proposes RKF-YOLO, a lightweight dual-task model based on improved YOLOv11n that integrates Rep-CSP structural optimization, knowledge-transfer-enhanced training (KTET), and Focal-CIoU loss. Compared with YOLOv11n, RKF-YOLO reduces parameters and FLOPs by 38.2% and 38.1%, respectively, while improving $\mathrm{mAP}@0.5$ and $\mathrm{mAP}@0.5:0.95$ by 0.6 and 1.1 percentage points for parking detection; for plate detection, Focal-CIoU improves $\mathrm{mAP}@0.5:0.95$ by 1.3 percentage points and contributes to a recognition accuracy of 95.7%. The unified framework uses a shared backbone and task-oriented detection heads to support vehicle-level illegal parking detection and license-plate-oriented localization. Rep-CSP enhances multi-scale feature representation, asymmetric channel reduction with feature compensation reduces redundant computation, and KTET improves convergence through optimizer and learning-rate migration. Deployment on RK3588 achieves 59.5 FPS for parking detection and 95.1% recognition accuracy, demonstrating real-time performance and practical applicability on resource-constrained edge devices. Full article

As robotic systems evolve toward large-scale distributed architectures and cloud-edge collaboration, communication middleware has become a critical infrastructure that impacts system real-time performance and scalability. The traditional Robot Operating System 1 (ROS 1) communication architecture, which relies on a centralized master node, has limitations in dynamic network environments. Robot Operating System 2 (ROS 2) achieves decentralized communication through the introduction of DDS. However, the single Data Distribution Service (DDS) mechanism remains inadequate for cross-network communication and high-performance local data exchange. Addressing the current issue in ROS communication research: the coexistence of multiple mechanisms without a unified analytical framework or guidance for selection. This paper systematically traces the evolution of the ROS communication architecture from centralized to distributed systems. It constructs a unified analytical framework covering two dimensions: communication models and data transmission paths. Crucially, to overcome the unreliability of cross-protocol comparisons based on heterogeneous literature, this paper designs and executes a set of unified benchmark experiments on a controlled testbed. These experiments systematically evaluate the performance of two mainstream DDS implementations (CycloneDDS and FastDDS) across five key metrics: latency, throughput, jitter, scalability, and packet loss rate under load. Additionally, a comprehensive comparative analysis of the performance of three transmission modes is conducted. Based on this comprehensive evaluation, this paper summarizes the performance characteristics of different mechanisms and further proposes an optimization-based middleware selection method for quantitative communication mechanism selection under different workload and application requirements. This paper provides a systematic reference for the design and optimization of ROS communication systems and offers guidance for promoting the application of multi-middleware collaborative architectures in robotic systems. Full article

Surface cracks on mandrel bars affect product quality and production stability in seamless steel pipe manufacturing. Existing vision-based methods mainly rely on bounding-box detection, which is insufficient for precise crack delineation and geometric characterization. This study proposes a lightweight segmentation framework for online mandrel bar crack inspection using grayscale industrial images. Based on YOLO11n-seg, the framework incorporates single-channel input adaptation, lightweight network reconfiguration, and crack-oriented feature enhancement to improve the extraction of weak, thin, and irregular cracks while reducing computational cost. A dedicated industrial dataset and a sample-balancing strategy are introduced to alleviate severe crack–background imbalance. Based on the predicted pixel-level masks, crack area, projected length, maximum width, and average width are calculated for online evaluation. Experimental results show that the proposed method achieves a mask mAP@0.5 of 88.5%, a false negative rate of 1.72%, and real-time inference at 204 FPS with 3.01 GFLOPs. Field deployment further demonstrates the effectiveness of the proposed framework for online crack inspection and geometric parameter calculation of mandrel bars. Full article

Effectively modeling software failure behavior is crucial for reliability assessment and planning of releases. However, many current software reliability growth models assume that failures are independent and fault detection mechanisms are simplified. However, these assumptions may not accurately represent real-world testing environments. This study introduces a novel Nonhomogeneous Poisson Process (NHPP)-based Software Reliability Growth Model (SRGM) that includes dependent failure behavior and exponentially decaying fault detection rates to better reflect the software debugging process. The proposed model was validated using real failure datasets and compared with 17 existing models. The performance of the model was assessed using various goodness-of-fit criteria, such as errors, prediction accuracy, and metrics based on information theory. To provide a more thorough evaluation, a multi-criteria decision-making approach was used to rank the competing models based on their overall performance. Furthermore, a one-at-a-time sensitivity analysis was conducted to examine how the initial values of the parameters affected the model’s behavior. These findings indicate that the sensitivity of the model to this parameter varies depending on the dataset used. The results indicate that the proposed model achieved superior performance across multiple evaluation criteria and consistently obtained the best overall ranking under the integrated multi-criteria framework. In Dataset 1, the proposed model achieved the best performance in most goodness-of-fit criteria, whereas in Dataset 2 it produced the best results across all twelve evaluation criteria. The results show that the proposed model offers improved or competitive performance compared to existing models and provides greater flexibility in capturing complex failure processes within software systems. Full article

Executing multi-UAV cooperative pursuit in complex public domains requires balancing interception efficiency with flight safety under strict micro-platform constraints. Existing planners often struggle with high computational overhead or lack kinodynamic adaptability in heterogeneous environments. To address this, we propose AC2F, a lightweight Adaptive Coarse-to-Fine hybrid framework featuring a bidirectional state-switching mechanism. The framework utilizes the Apollonius circle for efficient global guidance during the coarse phase, dynamically transitioning to a Dynamic Window Approach (DWA) upon detecting path oscillations or entering terminal capture zones. To ensure robustness, a dual-layer parameter paradigm integrates offline Bayesian optimization for globally optimal baselines with online real-time weight adaptation based on target distance. Extensive simulations show that AC2F effectively escapes local minima, such as urban-style U-shaped traps. Real-world suburban validation confirms an 86% capture rate with minimal computational overhead, demonstrating AC2F’s suitability for public domain protection and civil security. Full article

Background: Robot-assisted radical prostatectomy (RARP) is the predominant surgical approach for localized prostate cancer in high-volume centers worldwide. However, comprehensive real-world data describing complete institutional transition from open to robotic surgery remain limited. This study evaluated perioperative and early oncological outcomes of a contemporary RARP cohort and characterized the transition from open radical prostatectomy (ORP) to RARP in a European center. Methods: We analyzed 520 consecutive patients who underwent RARP between January 2023 and December 2025. Perioperative, pathological, and biochemical outcomes were assessed. Biochemical recurrence was defined as prostate-specific antigen ≥0.2 ng/mL. Institutional data from 2011 to 2025 were reviewed to evaluate procedural trends and the transition from ORP to RARP. Surgeon-specific and institutional learning curves were analyzed using operative time and linear regression models. Results: Following the introduction of robotic surgery in 2018, annual RARP volume increased from 37 procedures to 205 in 2025. Since 2023, RARP accounted for more than 99% of all radical prostatectomies. Median operative time decreased from 185 min in 2023 to 165 min in 2025, with consistent downward trends observed across all surgeons. Linear regression confirmed progressive improvement in operative efficiency, with learning rates ranging from −0.22 to −0.92 min per case. Estimated blood loss was minimal, no patients required transfusion, and major complications occurred in four patients (0.8%). Hospital stay decreased from 2 days to predominantly 1 day. During follow-up, 36 patients developed biochemical recurrence or PSA persistence. Biochemical recurrence-free survival differed significantly according to pathological stage (log-rank p < 0.001), with 24-month estimates of 93.7%, 91.5%, and 82.1% for pT2, pT3a, and pT3b disease, respectively. Conclusions: RARP provides favorable perioperative safety, minimal morbidity, and favorable early oncological outcomes in a high-volume setting. The complete institutional transition from ORP to RARP, together with demonstrated surgeon-specific and institutional learning effects, supports the feasibility and safety of implementing RARP as the institutional standard within a structured robotic program. Full article

Post-Traumatic Stress Disorder (PTSD) is a prevalent and costly mental health condition characterized by symptoms such as hyperarousal, avoidance, and re-experiencing. While machine learning (ML) approaches have shown promise in detecting PTSD-related physiological patterns, most validation efforts rely on computational metrics rather than real-world user perceptions. This study evaluates the perceived precision of a smartwatch-based ML tool designed to detect PTSD hyperarousal events using heart rate and activity data. The tool, previously developed using XGBoost 1.0.0, was deployed in a 21-day naturalistic study with 12 participants diagnosed with PTSD. Quantitative results showed a median perceived precision of 65.27%, with substantial variability across participants. A Mann–Kendall trend analysis revealed a significant increase in perceived precision over time for most participants, suggesting calibration of trust. Qualitative findings indicated high usability, general trust in the system, and acceptance of false positives, though concerns about notification design and battery life were noted. The results highlight the importance of incorporating user-centered, real-world validation into the evaluation of ML-based mental health monitoring tools. This work provides preliminary evidence supporting the feasibility of wearable-based PTSD monitoring and underscores the role of perceived precision in technology adoption and sustained use. Full article

The rapid proliferation of cloud data centers has intensified concerns over carbon emissions, energy efficiency, and sustainability. Virtual machine (VM) placement is a pivotal control lever, yet existing methods rarely couple carbon intensity signals with computationally tractable multi-objective optimization. In this paper, we propose CASO (Carbon-Aware Surrogate-Guided Optimization), a novel framework that integrates an online adaptive Radial Basis Function (RBF) surrogate model with a self-adaptive hybrid PSO-DE swarm optimizer for real-time VM placement in geo-distributed edge cloud environments. CASO simultaneously minimizes carbon emissions, energy consumption, SLA violation rate, and network latency under strict host capacity and Quality-of-Service (QoS) constraints. Three key innovations differentiate CASO: (i) an online surrogate update mechanism that refines fitness approximations incrementally as workload patterns evolve; (ii) a carbon intensity weighting scheme anchored to real-time Grid Emission Factor (GEF) signals; and (iii) an adaptive parameter controller that autonomously tunes swarm exploration–exploitation trade-offs without hand-crafting. Experiments on the publicly available Alibaba Cluster Trace (cluster-trace-v2026-GenAI) dataset within a CloudSim-Plus environment show that CASO reduces carbon emissions by up to 31.4%, energy consumption by 27.9%, and SLA violations by 18.8% compared to the strongest baseline while converging 3.8× faster than the strongest baseline (ADEDL). Full article

This paper presents Resilient Edge-IVA (Intelligent Video Analytics), an integrated framework designed to ensure real-time inference stability and high-speed embedding-based similarity search in resource-constrained edge computing environments. Conventional systems often face Quality of Experience (QoE) degradation caused by computational overhead and hardware-level bottlenecks. To address these challenges, this study proposes a “Whole-cycle” methodology employing a perception-driven, three-tier adaptive control algorithm. This algorithm dynamically modulates encoding parameters, such as resolution and bitrate, by utilizing real-time inference latency and CPU utilization as feedback signals. Furthermore, the framework incorporates an event-density-based Data Diet mechanism. This mechanism selectively adjusts video quality based on object detection results, preserving high-fidelity imagery for critical events while significantly reducing data volume during static intervals. The backend implements a hybrid storage architecture combining the Milvus vector database for CLIP-based high-dimensional visual embeddings with a PostgreSQL relational database for structured metadata. These systems are linked via a deterministic hash key to ensure data atomicity and facilitate high-speed, multi-dimensional embedding-based retrieval. Experimental evaluations conducted on a Raspberry Pi 5 and Hailo-8 NPU demonstrate that the proposed framework maintains a frame drop rate below 0.3% even under extreme workloads, providing a 13-fold improvement in operational stability over static configurations. The results also confirm a 54.2% reduction in total storage occupancy and a Hash Mapping Consistency (HMC) score of 0.89. These findings validate the framework’s effectiveness in reconciling real-time processing stability with storage efficiency. Building upon this baseline, future research will extend the framework to multi-class environments, targeting applications such as Intelligent Transport Systems (ITS). Full article