Search Results (4,570)

Indoor service and inspection robots need environment knowledge that is queryable, reliable, and maintainable as the site changes, yet open-vocabulary 3D pipelines propagate unverified labels downstream and build-once scene models offer no maintenance path. We present TLS-SMF, a semantic modeling framework that converts registered terrestrial laser scans into a confidence-aware semantic knowledge graph and keeps it current across repeated scans. TLS-SMF combines (i) a deterministic geometric backbone whose re-runs are byte-identical on the same host, (ii) multi-view vision–language-model verification with automatic escalation that retains unresolved objects as explicitly unverified, and (iii) identity-preserving incremental updates that apply confidence-gated node-level upserts to mint-once node identities. Across two operational configurations of one industrial room, escalation lifts verification accuracy from 77.8% to 88.9% and from 78.1% to 84.4%, with 94.4–100% unanimous-vote precision against confirmed owner ground truth; on the showroom benchmark, structurally gated querying answers 83.3% of questions versus 52.1% for an unverified Stage-A LLM baseline while blocking hallucination traps; in a controlled same-quality combined-edit re-scan, selective re-verification reduces VLM cost to 3.1% of a full rebuild; and a real three-epoch study detects 2/2 controlled physical changes with zero observed identity switches on strong reference correspondences, while re-segmentation still produces false-new/false-absent cases. A cross-floor cafeteria stress scene bounds domain transfer, and an owner-in-the-loop revision path keeps the single semantic store corrigible for robot-facing use; the same store drives a physical mobile manipulator, which completes three console-issued semantic missions with a mean arrival error of 0.30 m while phantom goals are refused with zero motion. Full article

To address the pressing demand for autonomous navigation in GNSS-denied environments, this paper systematically reviews the development status and future trends of multi-domain geomagnetic navigation and positioning technologies. First, from the perspectives of scalar and vector measurement systems, the technical characteristics of main-stream magnetometers, including optically pumped magnetometers, proton precession magnetometers, and fluxgate magnetometers, are comparatively analyzed, and the key parameters of several high-resolution global and regional geomagnetic field models are summarized. Then, geomagnetic navigation algorithms are classified into four categories, namely geomagnetic matching, geomagnetic filtering, bio-inspired navigation, and AI-assisted methods, and the core principles and technical features of each category are discussed in detail. On this basis, representative application scenarios in aviation, marine and underwater systems, and indoor environments are examined, with emphasis on their technical bottlenecks, representative experiments, and development trends. Finally, future directions are outlined from three aspects: high-precision magnetic data acquisition and high-resolution reference map construction, deep fusion of multi-source information and intelligent algorithm innovation, and engineering deployment and industrialization. This paper aims to provide theoretical references and technical support for the development of passive autonomous multi-domain navigation capabilities. Full article

Hempcrete is being recognized as an alternative bio-based insulation material for use in residential construction. A mixture of hemp fiber and hurd, lime, and water, this product can substitute fiberglass batt insulation and rigid foam insulation and reduce the carbon emissions associated with residential construction. While there are some studies on the thermal resistance of hempcrete and walls made with hempcrete, actual data on indoor environmental quality (IEQ) and energy performance of homes built with hempcrete are scarce. The research reported here presents an indoor environmental quality study based on measurements of the air inside a renovated house in New Castle, Pennsylvania, USA, that used hempcrete for walls. This paper also reports on the energy performance of the case study house based on energy modeling using the BEopt software tool using a comparative approach to show energy saving potential compared to the case of conventional batt insulation use. The project was initiated by DON Enterprises, Inc. to renovate a blighted house in New Castle, in an effort to demonstrate the feasibility of incorporating hempcrete into sustainable home building. This paper reports experimentally measured IEQ results and develops energy models and analyses of home energy use and discusses the evaluation of the energy efficiency of hempcrete walls compared with conventional fiberglass batt insulation. Full article

Air-source heat pumps suffer from the formation of frost on their outdoor coils while operating under cold and humid conditions, and the resulting periodic defrost cycles interrupt the heat supply and degrade both thermal comfort and efficiency. This study extends the hybrid air conditioner previously proposed by the authors, in which a ground heat exchanger is placed downstream of the outdoor heat exchanger, employing a shallow-buried water tank as a supplementary ground heat source to mitigate defrosting. A quasi-steady R32 vapor-compression cycle model was developed, coupled with lumped-capacitance thermal models of the room, the water tank, and the ground, including a frosting/defrosting model. Considering 2.2 kW heating at an outdoor temperature of +3.5 °C (with a relative humidity of 80%), the simulation predicted that the cumulative defrost time over 6 h would be reduced from 35 min (five events) to 0, the time required to reach the 20 °C set point would be shortened from 120 to 72 min, and the integrated coefficient of performance (COP) over 6 h would improve from 5.4 to 7.1. A preliminary one-hour winter heating experiment on a minimally modified commercial unit—one run per operating mode—showed that no defrosting occurred during hybrid operation, despite a higher outdoor relative humidity (74% vs. 46%), whereas two defrost events (inferred from the indoor outlet air velocity) totaling 12% of the run occurred in conventional operation, and the average COP increased from 1.8 to 3.0. An additional simulation with the as-built parameters of the prototype (300 L tank, tank-side conductance of about 60 W/K) reproduced this qualitative pattern under the hybrid-run weather, although with a small, nonzero predicted frost accumulation that remained below the defrost threshold. Because the two runs were conducted under different ambient conditions and the model was not calibrated against the tested system, the experiment provides qualitative rather than quantitative support for the simulated behavior; nevertheless, the simulation and the experiment consistently indicated the same relative advantage of the hybrid system, namely, the absence of defrosting and a higher COP. Future work will address repeated and extended experiments under matched conditions, model calibration, the optimization of the tank’s geometry and capacity, and the necessity of water agitation inside the tank. Full article

Wi-Fi channel state information (CSI) supports human activity recognition from motion-induced changes in indoor propagation, yet device, environment, and user changes perturb both channel responses and subcarrier relations. We present a model that aligns valid subcarriers across inputs and combines CSI response features with a multiresolution graph. Temporal decomposition produces residual components at multiple resolutions and a smooth component. The graph propagates information along positive and negative relations between subcarriers within each resolution and uses signal energy to guide propagation across resolutions. A gate incorporates the graph representation into the response classifier. A controlled multipath study further tests the physical interpretation of signed amplitude correlations. Experiments on CSI-Bench yield weighted-F1 scores of 95.96%, 50.16%, 44.08%, and 52.19% on the four protocols, with a cross-domain mean of 48.81%, exceeding the strongest of ten locally reevaluated baselines by 4.84 points. Full article

This study proposes a lightweight domain-assisted deep learning framework for binary human presence detection using ultra-wideband (UWB) radar. The proposed methodology processes raw UWB radar signals through statistically screened, physics-grounded signal features including Fast Fourier Transform (FFT)-based frequency-domain statistics and Hilbert Transform (HT)-derived envelope statistics which are selected via a per-subject Cohen’s d screening step and stacked as auxiliary input channels alongside the raw signal for a lightweight two-dimensional convolutional neural network (2D-CNN). A cross-subject evaluation protocol (train-on-one-subject, test-on-the-other) is adopted to assess generalization across individuals rather than relying on a pooled, sample-level split. Among the candidate features, a Frequency Standard Deviation (FSTD) is shown to match or exceed the performance of every multi-feature combination tested, indicating that targeted feature selection is more consequential than input fusion for this task. To further improve deployment efficiency, post-training INT8 quantization is applied, reducing the model to approximately 23 KB while preserving classification performance for quantization-robust configurations. Hardware-in-the-loop benchmarking on the STEdgeAI platform indicates on-device inference times ranging from approximately 0.88 ms on AI-enabled STM32N6 hardware to 117–130 ms on STM32H7-class microcontrollers; these figures reflect model inference only and exclude radar acquisition and preprocessing time. Experiments are conducted on a two-subject (one male, one female) indoor dataset; the reported cross-subject results are presented as a relative comparison across feature and quantization configurations rather than as an estimate of population-level generalization. The findings nonetheless illustrate the feasibility of combining principled feature selection with quantization-aware, hardware-validated deployment on embedded artificial intelligence (AI) platforms. Full article

Pepper suffers from bacterial leaf spot and yellow leaf curl, which substantially reduce crop yield and fruit quality. Traditional manual diagnosis suffers from delayed response, subjective bias and heavy labor consumption, while existing intelligent detection pipelines lack standardized preprocessing workflows, quantitative feature screening and systematic model comparison. To fill these research gaps, we built a pepper leaf dataset with 1260 samples (healthy, bacterial spot, yellow leaf curl). Three segmentation algorithms (Lab b-channel, RGB super-green, Otsu-ACWE) were quantitatively assessed to select the optimal preprocessing scheme. We extracted 32 fused visual features (27 RGB/HSV/Lab color moments + five gray-level co-occurrence matrix (GLCM) texture metrics) and adopted a random-forest classifier to eliminate seven low-contribution redundant features, retaining 25 discriminative variables. Three representative models, namely convolutional neural network (CNN), logistic regression (LR), and genetic-algorithm-optimized back-propagation neural network (GA-BP), were constructed for parallel comparison via 20 independent repeated trials, with accuracy, precision, recall, F1-score and area under the receiver operating characteristic curve (AUC) as evaluation indicators. The results verified that Lab b-channel segmentation achieved superior background separation and intact lesion edge retention. CNN yielded the best performance, with an average test accuracy of 97.67% and an average AUC of 0.999, accompanied by minimal metric standard deviations and outstanding stability. LR exhibits low computational cost and fast training, which is promising for applications with limited computing resources. In contrast, GA-BP shows weak nonlinear fitting ability and severe prediction fluctuations, making it unsuitable for high-precision diagnosis. This study proposes a standardized experimental framework to offer theoretical guidance and algorithmic references for intelligent vegetable leaf disease identification. All experiments were conducted on a dataset collected under standardized indoor single-illumination conditions; therefore, the conclusions of this study are only applicable to such controlled scenarios. Full article

With the development of intelligent manufacturing and smart logistics, automated guided vehicles (AGVs) are gradually expanding from traditional indoor environments to factory roads, logistics parks, industrial parks, and other outdoor road environments. However, low-visibility conditions, such as fog, rain, snow, and sandstorms, can degrade image quality, weaken object boundaries, and reduce the accuracy of visual perception systems. To address these challenges, an improved lightweight object detection model based on YOLOv11, named DMSM-YOLO, is proposed. To handle blurred boundaries, weak target features, and background interference in low-visibility images, the model is optimized from feature extraction, feature fusion, and prediction refinement. First, EPConv is designed to enhance directional contour interaction and improve blurred-boundary representation. Second, C3k2-MDFI is constructed to strengthen multi-scale weak-target representation. Third, MLCA is introduced to recalibrate low-contrast fused features by combining local spatial details and global contextual information. Finally, SMDetect is developed to improve localization accuracy and confidence estimation for blurred and occluded objects. Experimental results show that the proposed model achieves mAP@0.5 values of 54.71% and 71.67% on the DAWN and RTTS datasets, respectively. Compared with YOLOv11n, the mAP@0.5 is improved by 5.36 and 3.26 percentage points, respectively, while the model maintains a compact size of 2.56 M parameters and 7.1 GFLOPs. Additional experiments further verify the effectiveness of the proposed method under low-light conditions. The proposed model improves detection accuracy and robustness under low-visibility conditions while maintaining lightweight characteristics, demonstrating its potential for AGV visual perception. Full article

In the northwest region of China, collapsible loess is widely distributed, and its special geological characteristics have brought numerous challenges to the construction of foundation engineering. As a new type of foundation form, the research on long and short pile foundations in the collapsible loess area of the northwest is still in the initial and exploratory stage. In-depth exploration of the bearing characteristics of long and short pile foundations in the loess area under the water immersion state is of crucial significance for the safety and stability of engineering construction in this area. This paper adopts the effective stress method, fully considers the important factor of pore water pressure, and combines it with the load transfer method to construct a calculation method for negative skin friction of long and short pile foundations in the collapsible loess area. Through theoretical derivation and analysis, it can be known that the calculation method established has many unique features. It explicitly defines the soil surrounding the pile as unsaturated soil, on the basis of which the effective stress method and the load transfer method are organically integrated, and then a segmented curve model for calculating the negative skin friction of long and short piles under the water immersion state was established. The advantage of this model lies in its comprehensive and in-depth consideration of the influence of multiple factors (including pore water pressure, effective stress, and pile-soil relative displacement) on the mobilization of skin friction. To further verify the accuracy and reliability of this calculation method, the indoor model tests were conducted: the axial force, the lateral friction resistance, and the vertical displacement of pile foundation were compared. The results show that there is a good agreement between the theoretical and experimental results, proving that the calculation method proposed in this paper is reasonable and feasible. The achievement of this research result can provide a valuable reference basis for the pile foundation design work in similar collapsible loess areas, help improve the quality and safety of foundation engineering construction in these areas, and promote the further development and progress of engineering technology in related fields. Full article

Phosphorus (P) leaching from dryland from purple soils poses a significant risk to water quality, yet effective mitigation strategies and their underlying microbial mechanisms remain poorly understood. This study aimed to evaluate the efficacy of biochar (B), a silicon-based conditioner (Si), and their combination (BSi) in controlling P leaching, hypothesizing that B would immobilize P while Si would mobilize it. The indoor soil column leaching experiments were conducted with four treatments (CK, B, Si, BSi), measuring leachate P fractions and soil P forms, and employed metagenomic sequencing combined with partial least-squares path modeling (PLS-PM) and Bayesian structural equation modeling (BSEM) to explore microbial functional mechanisms. Results showed that B alone reduced cumulative leaching of inorganic P (IP), organic P (OP), and total P (TP) by a range of 5.4–6.3%, while increasing available phosphorus (Olsen-P) by 39.4% in the surface layer. Si and BSi promoted leaching, with BSi reducing available P sharply, despite raising TP. Metagenomic analysis revealed that B suppressed subsurface IP solubilization genes (e.g., gcd, ppx) and optimized OP mineralization, whereas Si inhibited mineralization via reducing key microbial taxa. BSEM further identified water-soluble P (Water-P) and total nitrogen (TN) as direct positive drivers of inorganic P dissolution. Collectively, the key biological mechanisms for leaching reduction involve inhibiting subsurface IP solubilization, optimizing surface OP mineralization, and enhancing P transport/starvation responses. Collectively, biochar applied alone offers the optimal balance between P retention and crop-available P supply in dryland purple soils, and provides mechanistic insights—through functional gene profiling—that can inform the design of more sustainable P fertilization and leaching control practices. Full article

Administrative labels such as “organic” or “pasture-fed” animal food products inadequately capture the biological complexity of dairy production systems and consequently fail to differentiate potentially meaningful associations with nutritional quality, environmental impact and consumers’ health. We developed a functional classification framework based on three measurable traits: (1) botanical diversity, (2) concentrate dependence, and (3) input intensity, and applied it to six observational and/or experimental studies from tropical Mexico (n = 109 farms across Yucatán, Chiapas, and Colima federal states). Animal production systems were categorized as: (a) biodiverse grazing (silvopastoral, organic, or traditional grazing; >15 plant species or 8–14 species with concentrate < 25% dry matter intake and no synthetic inputs) or (b) conventional (monoculture/indoor systems; <5 plant species or 6–10 species with concentrate >40% with a use of agrochemicals). Biodiverse grazing systems produced lower daily milk yields (6.1 vs. 7.7 L d⁻¹; p < 0.0001) but yielded milk with more favorable cardiovascular lipid indices: atherogenic index (AI: 1.6–2.2 vs. 1.9–2.5), thrombogenic index (TI: 2.3–2.7 vs. 2.3–3.1), and health promotion index (HPI: 0.46–0.67 vs. 0.40–0.54). Artisan cheese from biodiverse systems contained substantially higher concentrations of tocopherols (127 vs. 77 mg 100 g⁻¹ DM) and monoterpenes (460–475 vs. 111–126 ng kg⁻¹). An intensive silvopastoral system reduced enteric methane emissions by 18.2% (376 vs. 460 g d⁻¹; p < 0.05) and by 21.3% per unit dry matter intake (24.0 vs. 30.5 g kg⁻¹ DMI; p < 0.0001). Forage nutritional quality (crude protein, neutral detergent fiber, acid detergent fiber) varied by species and season, with these parameters influencing estimated methane emissions. In a C57BL/6 murine model of high-fat diet-induced obesity, milk from biodiverse silvopastoral systems attenuated body weight gain (30.5 vs. 36.8 g; p < 0.05), preserved glucose tolerance (AUC: 32,616 vs. 45,248), nearly eliminated hepatic macrovesicular steatosis, and promoted smaller adipocyte size (1863 vs. 3880 µm²; p < 0.05) compared with conventional milk. These findings suggest that functional ecological traits provide a useful framework for evaluating dairy systems, with biodiverse grazing offering a potential convergence of environmental sustainability and preclinical health benefits. Nevertheless, external validation across diverse tropical regions and direct human intervention trials remain necessary to establish generalizability, and the associations identified should be considered hypothesis-generating rather than confirmatory. Full article

Maintaining stable temperature and relative humidity (RH) is essential for edible fungi cultivation, yet indoor climates exhibit nonlinear, coupled thermal–moisture dynamics that can degrade disturbance rejection and increase energy use. This study proposes a reproducible lumped-parameter temperature–humidity model and evaluates a dual-loop (SISO × 2) fuzzy-PID regulation strategy in MATLAB/Simulink under standardized simulation scenarios. The model formulates energy and moisture conservation using physically interpretable parameters (air mass, ventilation exchange, heat transfer, and actuator limits), with RH obtained from a psychrometric transformation of humidity ratio. Three benchmark tests are designed for repeatable assessment: set-point tracking, step disturbances (±2 °C and ±5% RH), and periodic disturbances. A Simulated Growth Indicator (SGI) is introduced as a phenomenological model representing potential growth trends under controlled temperature and humidity, rather than actual measured crop yield. The fuzzy-PID strategy is compared with conventional PID and a no-control baseline using unified performance metrics (steady-state deviation, recovery/settling behavior, integral error) and actuator energy consumption computed from explicit power models. Results show that fuzzy-PID achieves faster recovery and lower error accumulation than PID under identical disturbances while reducing total energy consumption (e.g., 5.7 kWh vs. 6.7 kWh in a representative case). Although the plant is dynamically coupled, the controller implementation remains a practical dual-loop structure; the presented framework therefore serves as a reproducible simulation benchmark for controller comparison in high-humidity cultivation stages. Full article

Modern building infrastructures are becoming increasingly complex, creating a need for intuitive indoor navigation systems that can assist users in unfamiliar environments. Augmented Reality (AR) has emerged as a promising solution by providing spatially contextual guidance directly within the user’s field of view. However, many AR indoor navigation systems rely on manually constructed 3D environments, a development process that is time-consuming and prone to spatial inconsistencies with the real-world environment. This study presents a comparative evaluation of two environment creation workflows for AR indoor navigation development: a traditional manual 3D modeling approach and an automated cloud-based spatial mapping workflow using the Immersal SDK. A counterbalanced within-subject experiment was conducted with 48 participants, each of whom completed equivalent indoor navigation development tasks using both workflows in a real-world campus building environment. The development process was divided into three stages: environment acquisition, environment generation, and system integration. Development efficiency was evaluated using stage-based development time measurements, while perceived workload was assessed using the NASA Task Load Index (NASA-TLX). Statistical analysis was performed using repeated-measures analysis to compare workflow performance across development stages. Results show that the automated workflow significantly reduced overall development time by approximately 38% compared to the manual modeling approach, with the most substantial time reductions occurring during the environment acquisition and environment generation stages. NASA-TLX results indicate an approximately 31% reduction in overall perceived workload. Descriptively, the automated workflow had lower mental-demand and effort scores but a higher physical-demand score. A separate researcher-conducted spatial validation of one implementation per workflow showed a higher mean three-dimensional positional error for the automated implementation (22.47 cm) than for the manual implementation (19.14 cm), with a mean paired difference of 3.33 cm across 13 anchor locations. These findings indicate that automated spatial mapping can substantially improve development efficiency and reduce overall perceived workload, while introducing trade-offs in physical demand and spatial alignment accuracy relative to manual environment reconstruction. Full article

Nowadays, speakers are essential components for message delivery in indoor environments, including lectures, public addresses, and emergency announcements. Their physical placement should ensure adequate direct-sound audibility across occupant service areas while maintaining installation feasibility. A digital twin is a technology that allows an infrastructure layout to be designed and evaluated virtually on a computer before physical installation by reconstructing an indoor environment as a 3D model. In previous studies, we have proposed a method to reconstruct a 3D indoor model of an indoor environment from its ${360}^{\circ }$ panoramic images using 3D Gaussian Splatting (3DGS) and a 3D point cloud, and applied it to surveillance camera placement. In this paper, we propose a digital twin-based speaker placement planning tool for indoor environments by generalizing the previous method to direct-sound acoustic simulation. This tool consists of four stages: (1) reconstructing a 3D indoor model from ${360}^{\circ }$ panoramic images and extracting floor and desk receiver surfaces, (2) assigning the initial speaker budget based on the reconstructed floor area, (3) determining speaker mounting positions on valid ceiling regions using K-means spatial clustering under obstacle and boundary constraints, and (4) simulating broadband direct-sound sound pressure level (SPL) across floor and desk receiver surfaces. For evaluation, the proposed tool was deployed across three real-world indoor scenarios: a basketball hall ($32.15 \mathrm{m}\times 21.99 \mathrm{m}$), a furnished office ($7.17 \mathrm{m}\times 6.17 \mathrm{m}$), and a non-convex L-shaped office ($23.00 {\mathrm{m}}^2$). The experimental results showed that in each scenario, the generated layout achieved complete direct-sound audibility compliance across all sampled physical test locations (≥60 dB floor/≥65 dB desk) with zero detected hotspots exceeding $85 \mathrm{dB}$, maintaining SPL values between $66.91 \mathrm{dB}$ and $77.88 \mathrm{dB}$. An on-site physical measurement campaign confirmed that the generated layouts satisfy target audibility thresholds under real room conditions, with mean absolute errors between $1.78 \mathrm{dB}$ and $2.41 \mathrm{dB}$. These results confirm the practical utility of our approach as an initial geometry-driven planning tool for indoor audio infrastructure deployment. Full article

Against the backdrop of the global energy transition and carbon reduction in the building sector, this study proposes a photovoltaic solar chimney coupled with an earth–air heat exchanger (PVSC-EAHE) system to integrate passive ventilation, thermal pre-treatment, and photovoltaic power generation in cold and arid regions. A coupled numerical model is presented using MATLAB 7.2 and TRNSYS 16.0, with user-defined models for the earth–air heat exchanger, solar chimney, and photovoltaic system implemented as TRNSYS-callable modules. The novelty of the study lies in the systematic year-round investigation of five inclination angles of photovoltaic (PV) panel (20°, 30°, 40°, 50°, and 60°) and their effects on the coupled thermal, ventilation, and electricity-generation performance of the system under different seasonal conditions. An 8760 h annual dynamic simulation was conducted using typical meteorological data for Hami, Xinjiang, China, to evaluate system performance and the effect of PV inclination angles ranging from 20° to 60°. The results show that, compared with a conventional building without the coupled system, the PVSC-EAHE system reduces the average indoor summer temperature by approximately 4 °C, maintaining temperatures below 28 °C, while increasing winter indoor temperatures by up to 8 °C. The system also provides an average fresh-air supply of 132 m³/h and generates 2180.1 kWh of electricity annually. The inclination analysis reveals a season-dependent trade-off: higher inclination angles are more favorable for solar radiation capture and ventilation during winter, whereas lower angles are comparatively more favorable during summer. The inclination angle has a relatively limited effect on indoor thermal conditions, while increasing it from 20° to 60° raises the maximum instantaneous PV output from 198 to 220 W/m². Overall, the results demonstrate the potential of the PVSC–EAHE system to improve indoor thermal conditions, enhance natural ventilation, and generate renewable electricity in cold and arid climates. As annual heating and cooling consumption was not explicitly calculated, the observed performance improvements indicate the potential to reduce building energy demand rather than quantitatively demonstrating specific annual energy savings. Full article