Search Results (7,543)

This study presents a practical approach to developing a prototype outdoor cleaning robot with reduced hardware and software complexity. While indoor cleaning robots are widely adopted, their outdoor counterparts remain costly and less accessible due to unstructured environments and increased system complexity. To address this, we retrofitted an off-the-shelf manual sweeper with minimal hardware: a single 3D LiDAR and a compact onboard computer running an open-source autonomy software stack. This LiDAR-centric design reduces hardware and software integration complexity while using the integrated IMU of the LiDAR for motion compensation. We focus on adapting three key autonomy modules, mapping, localization, and planning, for semi-structured park environments characterized by dense vegetation and GNSS-denied conditions. A consistent and loop-closed pointcloud map was generated using a LiDAR SLAM framework. Map-based pose initialization and LiDAR scan matching were evaluated under the tested conditions. Bidirectional path-following behavior was demonstrated through lanelet-based planning adapted for shared narrow paths. A field trial demonstrated autonomous execution of the predefined cleaning route, with a loop-closure residual of 0.072 m, low covariance-derived lateral uncertainty, and a mean deviation from the lanelet-centerline reference trajectory of approximately 0.14 m. These results demonstrate the feasibility of a low-complexity LiDAR-centric autonomy architecture for the tested outdoor cleaning scenario. Full article

Structural deformation of large-scale phased-array antennas under complex service environments seriously degrades the electromagnetic performance of radar systems. It is critical to achieve high-precision full-field deformation reconstruction of antennas with limited sensing data. In this paper, a multi-fidelity data-driven generative adversarial network framework is proposed to predict the full-field displacement of antennas from discrete strains at limited measuring points. The proposed model is validated for antenna surface reconstruction under multiple loading conditions. Large-scale datasets generated by finite element analysis are adopted for low-fidelity simulation pre-training, and a small amount of measured fiber Bragg grating (FBG) data are utilized for fine-tuning and calibration. Experimental validation was conducted under three independent loading conditions. The proposed method achieved RMSE values of 1.05 ± 0.03, 1.28 ± 0.03, and 1.54 ± 0.04 mm, respectively, with an average RMSE of 1.29 ± 0.035 mm. Compared with the SS-MLP regression model, DeepONet baseline model, and the multi-fidelity generator without the conditional discriminator, the average RMSE was reduced by approximately 36.4%, 22.6%, and 25.4%, respectively, demonstrating improved reconstruction accuracy and adaptability to different structural response patterns. Full article

Semantic segmentation enables pixel-level scene understanding, and deploying it on resource-constrained platforms such as unmanned aerial vehicles and embedded sensing devices has become increasingly important. In complex scenes, however, objects with substantial scale variation and heterogeneous morphology often lead to missed small objects, discontinuities along slender structures, and incomplete boundaries. To address the challenges, this paper proposes a lightweight semantic segmentation network, named MAF-Net. The network adopts MobileNet as the backbone and enhances the representation of objects with different scales and morphological characteristics through multi-scale feature extraction and top-down feature fusion, while maintaining low model complexity. To further accommodate variations in object scale and structure, a Morphology-Adaptive Kernel Decoder (MAKD) is designed. The module introduces large-kernel, small-kernel, horizontal-strip, and vertical-strip convolution branches, and adaptively fuses the multi-branch features through a gating mechanism, thereby providing complementary contextual representations with different receptive-field scales and directional sensitivities. In addition, a Triple Feature Fusion Refinement Module (TFFRM) is constructed to further refine the initial segmentation results by integrating shallow spatial details, learned boundary-aware responses, and high-level semantic features, thereby alleviating problems such as discontinuities in slender structures and blurred object boundaries. Experimental results show that the proposed method achieves favorable overall segmentation performance on the COCO2017-derived ball-sports dataset, UDD6, and WHDLD, while maintaining a relatively small parameter count and high measured inference speed. Full article

Chinese food safety standards and scientific literature contain quantitative information relevant to food safety supervision and risk analysis. However, this information is dispersed across natural-language provisions and complex tables, making manual extraction labor-intensive and prone to omissions and incorrect associations. This study proposes a large language model (LLM)-based framework that combines few-shot prompting with Self-Refine to extract “food object–indicator/factor–numerical expression” triples. A corpus comprising 149 Chinese national food safety standards and 85 Chinese scientific articles was constructed, and the evaluation used 1031 reference triples from 27 documents. Across four runs, Qwen-Plus with Self-Refine achieved mean precision, recall, and F1 scores of 0.8690, 0.8041, and 0.8353 under strict matching and 0.8737, 0.8084, and 0.8398 under similarity-based matching. Strict-match precision and F1 increased by 0.0421 and 0.0207, respectively, while recall changed little. The F1 improvement remained positive across the tested similarity thresholds and field weights. Without additional model training, the proposed framework produces structured quantitative food safety information that can support digital archiving of food safety documents, information retrieval, and knowledge graph construction. Full article

This study reconstructs long-distance trade networks within the hyper-arid Arabian Peninsula by evaluating the dynamic interactions between human activities and geo-environmental constraints. The primary objective is to delineate the optimal trajectory of the ancient Incense Trade Route from Qaryat al-Faw to Al-Yamama utilizing Least-Cost Path (LCP) analysis within a GIS environment. Applying a hydro-archeological framework, relative groundwater potential is conceptualized as a critical environmental constraint governing caravan routing and transit station spacing. Methodologically, an Analytical Hierarchy Process (AHP) and Weighted Overlay Analysis (WOA) within a Multi-Criteria Decision Making (MCDM) model integrated six hydro-topographic and climatic parameters—lithology, slope, drainage density, soil, elevation, and precipitation—to generate a Groundwater Potential Zone (GWPZ) map. A multi-parametric friction surface was then constructed by coupling slope, elevation, and lithology with proximity to wadi channels (Distance to Wadis) as a dedicated hydrographic friction parameter, simulating caravan mobility across least-resistance conduits. The geospatial simulation generated a continuous transport corridor spanning 605.92 km, traversing minimal topographic resistance and maximizing correlation with hydrographic wadi networks. Spatial overlay analysis demonstrates a strong spatial alignment between ancient occupation centers and zones of favorable hydrological potential: primary macro-regional stations such as Al-Yamama and Qusayrat Aad intersect Very High Groundwater Potential Zones, while Qaryat al-Faw and al-’Jiliyya fall within High potential zones, and Ayun Al-Aflaj aligns with a Moderate potential zone. Proximity analysis demonstrated high geographic accuracy, yielding a mean spatial distance deviation of only 2.78 km between the simulated path and documented archaeological stations. Furthermore, remote sensing analysis along the LCP corridor identified 158 distinct archeological features, predominantly diagnostic pendant stone structures and necropolis complexes morphologically identical to the monuments overlooking Qaryat al-Faw. This linear distribution of funerary footprints supports the spatial plausibility of the historical path, demonstrating that integrating multi-criteria GIS processing with hydrographic analysis offers a robust predictive framework for prioritizing future targeted field surveys. Full article

Under global climate change and rapid urbanization, urban heat island and dry island effects threaten residents’ thermal comfort and public health, making microclimate regulation by urban green spaces a core issue. However, the interaction between vertical structure and horizontal spatial type of plant communities remains unclear. This study selected Tianjin Water Park as the study area. Based on field temperature and humidity observations during the high-temperature period in summer 2024, ENVI-met was locally validated and used for multi-scenario numerical simulations of typical plant communities. Results showed that all vegetation structure types had cooling and humidifying effects. Tree–shrub–herb communities showed the strongest cooling and humidifying performance, while shrub–herb communities showed the weakest performance (p < 0.05). A higher degree of horizontal enclosure was associated with more stable temperature and humidity conditions. For cooling, an interaction-dependent effect was found between vertical structure and horizontal spatial type: the cooling benefit of greater vertical structural complexity increased with the degree of enclosure. These findings clarify the independent and interactive mechanisms by which plant community structure regulates microclimate. They also provide a scientific basis and quantitative reference for optimizing plant arrangement in urban parks. Full article

Artificial intelligence (AI)-based mineral prospectivity mapping (MPM) has rapidly advanced through the integration of deep learning, multisource geoscience data, and increasingly complex representation-learning approaches. However, the reliability of AI-derived prospectivity predictions remains a critical challenge because high predictive performance does not necessarily indicate geological validity, spatial transferability, or exploration usefulness. The key challenge is therefore not model complexity, but aligning geological knowledge, data representation, validation, and uncertainty assessment to ensure reliable interpretation. This review provides a structured critical synthesis of geological reliability in deep learning-based MPM, focusing on the methodological factors that influence whether AI outputs can support defensible exploration decisions. Rather than comparing algorithms solely by predictive accuracy, we examine the interconnected roles of geological representation, prediction support, label construction, negative-sample strategies, validation design, uncertainty quantification, explainability, transferability, and independent geological verification. Assessment of representative AI-based MPM studies reveals inconsistent reporting of these reliability dimensions, limiting reproducibility and meaningful comparison across studies. The review identifies recurring reliability bottlenecks, including spatial leakage caused by inappropriate validation strategies, uncertainty introduced during evidence-layer construction, ambiguity associated with unlabelled geological regions, and limitations in interpreting model explanations as geological causality. To organize these challenges, we propose a hierarchical evidence framework consisting of eight levels: methodological feasibility, internally validated predictive performance, spatial or temporal transferability, geological coherence, calibrated uncertainty, independent geological evidence, prospective field or drilling evidence, and demonstrated decision value. This framework clarifies the strength of claims that can be supported by different levels of evidence and provides a basis for transparent reporting and reliability assessment. Future progress in AI-based MPM will require integration of predictive capability with geological knowledge, robust validation, uncertainty-aware modelling, interpretable predictions, and independent geological verification. The proposed framework aims to support the development of more reproducible, transferable, and decision-relevant AI-assisted mineral exploration workflows. Full article

The Malavedan language is a severely endangered Dravidian tribal tongue spoken in the Kollam, Kottayam, and Pathanamthitta districts of Kerala, India. While historically overshadowed by its dominant neighbour, Malayalam, current field documentation (2025) reveals a pronominal and address system that has resisted the hierarchical complexities characteristic of the broader region. This paper documents that system and frames it as a bidirectional comparative study: it examines not only what Malavedan lacks relative to Malayalam, but what Malavedan possesses that Malayalam does not. Malayalam encodes a graduated honorific system through its second-person pronominal distinctions, /n̪iː/, /n̪iŋŋaɭ/, and /t̪aːŋkaɭ/, and through third-person honorific forms /id̪d̪eːham/ and /ad̪d̪eːham/ that grammaticalize deference toward a referred-to third party. Malavedan has none of these forms. ‘Egalitarian’ is used here in a strictly grammatical sense, to mean that the pronominal paradigm does not rank the people it refers to; it does not mean that the community lacks deference, which it expresses through titles, kin terms and bodily conduct. What Malavedan has instead is a three-way proximal/distal/far-distal deictic contrast that Malayalam lacks entirely. The far-distal forms /t̪oːʋarə/ (that man, far) and /t̪oːt̪t̪uʋə/ (that woman, far) fuse the deictic distance marker /t̪oː-/ with adult gender information in a single form, a strategy that has no equivalent in Malayalam, whose pointing word /at̪aː/ ‘there’ carries no gender. When the referent is a child, for whom adult gender distinctions are inapplicable, the gender-neutral far-distal form /t̪oːt͡ʃt͡ʃə/ is used instead, with gender if needed carried by the accompanying noun. This adult/non-adult split within the far-distal demonstrative slot has no Malayalam parallel. Taken together, these distance contrasts point to a language that has invested its grammatical resources in locating people in physical space, where Malayalam has invested its own in ranking them socially. The structural gap produced by Malavedan’s egalitarian architecture generates a pragmatic clash when speakers interact with Malayalam-medium institutions, forcing reliance on borrowed ‘Sir’ and ‘Madam’ to navigate social hierarchies that the native system has no grammatical resources to express. Despite this structural flatness in the social domain, Malavedan possesses a sophisticated insider lexicon that performs crucial social work: coded referents for police (/kiːt̠t̠uʋaːjan/), exonyms for surrounding communities, an endonym (/t͡ʃaːd̪ikaːɾə/) that asserts collective identity on the community’s own terms, and ambivalent status terms, like /ɲon̪d̪alə/. Intergenerational transmission has effectively broken down, and children now operate almost entirely within the Malayalam framework. Documenting this shift is more than lexical salvage. It records an indigenous social map before the pressure of a dominant, deeply stratified language erases the pragmatic logic on which that map was built. Full article

Radio altimetry is important for sensing the near-ground altitude of airborne platforms operating at several-kilometre altitudes. In urban environments, multipath propagation and coherent superposition from buildings and ground surfaces can alter echo delay, energy, and waveform, thereby affecting downstream ranging and altitude-estimation decisions. Conventional echo models often cannot retain the geometric propagation delays, material properties, polarization transformations, and coherent phase relationships of individual scattering components in kilometer-scale three-dimensional scenes, limiting their ability to reconstruct physically consistent time-domain echoes. To address this, this study combines radio-altimetry echo simulation with electromagnetic simulation of urban environments and develops a finite-distance spherical-wave physical-optics method for modeling time-domain echoes in large-scale urban scenes. The method maps the scattering responses of urban surface facets to the received time-domain echo. Representative cases verify the accuracy of calculations of scattering amplitude, range attenuation, and two-way propagation delay. The results show that a uniform material assignment predominantly affects echo amplitude and energy while exerting only a limited influence on the delay structure. In contrast, variations in urban geometry alter the spatial locations and orientations of effective scattering facets and the coherence relationships among their complex fields, resulting in distinct main-peak positions, energy-delay distributions, and early- and late-arriving multipath structures. Full article

A comparative descriptive study was conducted to characterize the morphogenetic and physical properties of six soil profiles in the Kostanay region of northern Kazakhstan. This study aimed to differentiate the soil genetic horizons, assess their structural quality, and investigate their physical properties under distinct ecological matrices, including baseline zonal steppes, Amankaragay intra-zonal sandy pine forests, dynamic river floodplains, and agriculturally modified areas. Integrated field macro-morphological profiling and genetic diagnostic techniques were used to evaluate these properties. It was found that the soil cover in the area consists of ordinary chernozem, southern chernozem, dark chestnut soil, alluvial soil, and sandy sod-forest soil. Differences were identified in the thickness of the humus layer, soil structure, and depth of carbonates. The macro-morphological analysis revealed distinct structural and genetic features across the studied landscape matrices. The steppe zone had well-developed, granular-blocky profiles with carbonate effervescence that was depth-dependent (down to 140–150 cm) and localized solonetzic features. The Amankaragay forest sands had in situ weathering (brunic features) and deeply developed humic profiles. The Tobol River floodplain had a complex, stratified sequence with a redox-mottled, buried meadow horizon [A] at a depth of 63–68 cm. The results showed that the studied chernozem soils had favorable agrophysical properties and robust structural integrity. They also had typical humus profiles, which create a fundamental physical baseline for soil moisture retention and provide a foundation for future comprehensive chemical fertility assessments. Full article

In the present paper, a comprehensive review of methodologies for the optimal placement of measurement devices in chemical processes, commonly referred to as the sensor network design problem (SNDP), is presented. The SNDP is a combinatorial optimization problem of paramount importance in process design, monitoring, control, and data reconciliation, as sensor placement directly affects estimability, precision, reliability, and fault detectability. In addition to surveying solution strategies for SNDP, this review examines the closely related variable classification problem, which constitutes a fundamental building block in the systematic development of sensor networks. The evolution of the field is analyzed, highlighting early structural search approaches, subsequent optimization-based formulations relying on relaxation and decomposition techniques, and heuristic and metaheuristic strategies. The strengths and limitations of these methodological classes are discussed with respect to problem size and computational complexity. Moreover, this review identified a novel strategy to solve SNDPs based on machine learning algorithms. Finally, future research directions are outlined, emphasizing the growing need for scalable frameworks capable of addressing increasingly large and automated industrial systems. As computational capabilities continue to advance and process plants transition toward highly digitalized environments, research efforts devoted to efficient and robust SNDP methodologies are expected to intensify. In this context, Large Language Models (LLMs) can play a transformative role. Full article

Chlorogenic acid (CA) has received extensive attention in the field of improving the properties of starch. At present, few studies have focused on the influences of chlorogenic acid (CA) addition levels on the physicochemical properties of Chinese yam (Dioscorea opposita Thunb.) starch (CYS). Therefore, the effects of varying chlorogenic acid (CA) concentrations (3%, 5%, and 7%) on the structural characteristics, in vitro digestibility, and rheological properties of Chinese yam starch (CYS) and its CA-incorporated complexes was investigated. The incorporation of CA disrupted the double-helix structure of starch, while promoting the formation of double- and single-helical structures in amylose, as revealed by Fourier-transform infrared spectroscopy and Raman spectroscopy, respectively. This led to decreased crystallinity and the formation of a V-type crystalline structure. The interactions between CA and CYS molecules facilitated the construction of a more compact network architecture within starch granules, thereby decreasing its in vitro digestibility and significantly increasing resistant starch content. Moreover, CA enhanced the short-range ordered structure of starch, restricted granule swelling, strengthened shear-thinning behavior, and reduced gel strength and viscosity. In summary, the complexation between Chinese yam starch and chlorogenic acid provides a promising strategy to tailor the rheological properties and in vitro digestibility of starch, providing a promising candidate substrate for fabricating low-digestibility starchy functional food formulations. Full article

The efficacy of microbial inoculants in enhancing plant drought tolerance is context-dependent, varying with drought severity and the dynamic interactions within the plant–soil microbe system. We conducted a two-factor pot experiment using Achnatherum splendens under three moisture regimes (well-watered (control), moderate drought, and severe drought) and four inoculation treatments (sterile inoculum (control), Bacillus subtilis (BS), arbuscular mycorrhizal fungi (AMF), and their co-inoculation (MIX)). The resulting 12 treatment combinations were each replicated four times. Drought stress significantly reduced both shoot and root biomass, disrupted redox homeostasis by altering antioxidant enzyme activities and osmolyte accumulation, and induced pronounced shifts in rhizosphere microbial community composition and structure. Microbial inoculation modified osmolyte accumulation and antioxidant enzyme activity, with effects varying in magnitude and direction across moisture regimes and physiological indicators, and was generally associated with elevated soil nutrient availability and extracellular enzyme activities, though the magnitude of these effects was modulated by moisture regime. The MIX treatment showed the strongest positive responses for a subset of plant physiological and soil biochemical indicators. Bacterial and fungal co-occurrence networks exhibited divergent structural responses to inoculation and drought, with network complexity (e.g., connectivity and modularity) differentially shaped by both factors. Mantel tests and partial least squares path modeling revealed that under well-watered and moderate drought conditions, soil physicochemical properties were the primary drivers of microbial alpha and beta diversity; in contrast, under severe drought, osmotic adjustment emerged as the dominant physiological determinant of biomass maintenance. Collectively, these results reveal that the functional outcomes of microbial inoculation are contingent upon abiotic context, particularly drought intensity. This study provides a mechanistic basis and preliminary empirical evidence for subsequent field validation and further development of moisture-matched microbial intervention strategies for the ecological restoration of drought-vulnerable A. splendens ecosystems. Full article

Finite element analysis is essential for ship structural design because it enables accurate assessment of structural strength, fatigue, and vibration performance. However, mesh generation for large-scale hull structures remains challenging due to the trade-off between computational efficiency and local mesh resolution. This study proposes a semi-automatic hybrid coarse–fine mesh generation framework based on precise geometric models for local refinement. Users mark critical regions during geometric modeling; the framework then processes them automatically through a rule-driven strategy incorporating Common Structural Rules (CSR) requirements, while coarse meshes are retained in non-critical areas. A subdivision approach combining trisection with proportional and non-proportional transition patterns is developed to achieve smooth element-size variation between different mesh regions. For complex structural features, including stiffener-frame cut-outs and openings, an enhanced cross-field method is introduced to generate high-quality quadrilateral meshes while preserving geometric characteristics. A boundary simplification strategy is further developed to avoid mesh degradation near complex boundaries. Validation using representative transverse and bottom-side tank sections of a very large crude carrier (VLCC) with a deadweight tonnage (DWT) of 300,000 tonnes demonstrates the applicability of the proposed framework to local mesh refinement. The final refinement zones meet the specified mesh-density and geometric-quality criteria, whereas some transition-zone elements fall outside the adopted quality limits. Validation addresses local refinement results and geometric mesh quality rather than structural response. The proposed framework supports semi-automatic local mesh refinement of large-scale hull structures. Full article

The theoretical description of phase transitions in condensed matter represents one of the most profound and enduring challenges in modern physics, spanning more than a century of intellectual development. From the phenomenological models of van der Waals in 1873 and Landau in 1937, through the statistical-mechanical rigor of integral equation theories of Ornstein and Zernike in 1914, Percus and Yevick in 1958, to Wilson’s renormalization group in 1971 and contemporary spectral methods such as the Self-Consistent Spectral Potential Mean-Field (SC-SPMF) method of Villani, the theoretical landscape has evolved through successive refinements that have progressively deepened our understanding of critical phenomena, metastability, and universality. Despite these remarkable advances, the prediction of phase transition temperatures, the characterization of critical behavior, and the understanding of metastability and hysteresis remain active areas of research, particularly for first-order transitions such as melting and crystallization. This review develops a central thesis: the difficulty of describing phase transitions in condensed matter arises from the intrinsically irreducible complexity of N-body systems. The behavior of a system of N interacting particles cannot, in general, be reduced to the behavior of N independent two-body systems. This is the essence of the N-body problem: the interactions between particles create collective effects that cannot be understood by considering pairwise interactions in isolation. The irreducible nature of this problem manifests itself in the hierarchy of correlation functions, where the n-body distribution function depends on the (n + 1)-body distribution function, leading to an infinite hierarchy that cannot be closed without approximations. The SC-SPMF method resolves the complexity of the problem through a spectral reduction that transforms the irreducible N-body problem into a self-consistent two-body problem in a mean field. The spectral criterion for phase transitions introduced by the SC-SPMF method represents a fundamental departure from traditional approaches: the transition is marked by a change in the curvature of the first eigenvalue μ₁ (T). This criterion does not rely on arbitrary thresholds but on a qualitative change in the spectral response of the system, reflecting a fundamental change in the nature of correlations. The review is organized into three main parts. The first establishes the historical and conceptual foundations, tracing the evolution of theoretical approaches from van der Waals to the renormalization group, examining the successes and limitations of each approach, and identifying the common thread of approximate closures and functionals that has motivated the development of alternative methods. The second addresses the fundamental problem of complexity in N-body systems, exploring the BBGKY hierarchy, the irreducibility of many-body correlations, and the failure of pairwise reduction; it then presents the spectral mean-field approach as a rigorous solution to this problem, demonstrating how the eigenvalue spectrum of the Fisher operator encodes the collective behavior of the system. The third provides a comprehensive account of the SC-SPMF method. The SC-SPMF method offers several distinct advantages over existing approaches: it requires no approximate closures (unlike integral equation theories), no excess free energy functionals (unlike density functional theory), and no extensive computational resources (unlike molecular dynamics or Monte Carlo simulations). The coordination shell structure emerges self-consistently from the spectral decomposition, and the method naturally captures hysteresis and metastability because the self-consistent cycle provides stable solutions as long as the Fisher equation yields negative eigenvalues. The predictive power of the method is demonstrated by its ability to reproduce experimental structural parameters and to reveal the path-dependent universality that has not been observed in previous theoretical studies. Full article