Search Results (777)

Deep learning surrogate models trained with mean-squared-error loss produce statistically accurate but physically unconstrained flood predictions: water may flow uphill, appear spontaneously, or smooth over street-level corridors. In this work, a physics-informed training framework is developed for CNN-LSTM models that predict urban flood depths at 15 min intervals over a $128\times 128$ spatial grid. Three differentiable penalty terms are embedded directly into the loss function: (i) a gravity loss that penalizes depth increases against the water-surface-elevation gradient, (ii) a continuity loss enforcing local mass conservation with rainfall-adaptive thresholds, and (iii) a topography-aware false-alarm penalty modulated by the topographic wetness index (TWI). The framework is evaluated on the Norfolk, Virginia, flood dataset spanning two major storm events (August 2017 and September 2022) comprising 300 samples, with all variants trained on identical splits and robustness assessed over repeated random splits and leave-one-storm-out tests. A road-proximal evaluation restricted to a TWI-derived street mask quantifies street-level skill. The physics-constrained model achieves near-zero gravity violations (∼10⁻⁶) and the highest street-channel recall (0.77 ± 0.09 versus 0.44 ± 0.10 for the unconstrained baseline), the capability most relevant to downstream traffic routing, and its recall advantage more than doubles on a held-out storm, while a uniform false-alarm variant attains 16% lower mean absolute error but suppresses street recall to 0.25. The proposed TWI-modulated penalty reconciles this trade-off: it improves upon the uniform variant on every metric measured, recovering 60% higher street recall at the lowest MAE among all constrained variants and the best street-level F1 score. These results expose a fundamental tension between aggregate pixel-level error metrics and application-specific physical plausibility, and demonstrate that terrain-aware loss modulation offers a principled resolution. Full article

Learning transferable region embeddings is fundamental to urban computing, supporting applications from economic forecasting to public safety. Existing multi-view fusion methods predominantly rely on coarse-grained fusion, applying uniform weights across an entire city or task, thereby neglecting spatial heterogeneity—the fact that the dominant view driving a region’s function varies significantly across regions. To address this, we propose RAMEN, a two-stage framework for Region-adaptive Mixture of Ego-Networks learning. In the pre-training stage, a unified spatially aware Transformer distills universal urban semantics. In the fine-grained adaptation stage, a Mixture of Ego-Networks (MoEN) module employs a region-adaptive gating mechanism to dynamically allocate exclusive view weights for each region. A Region Ego-aware Spatial Transformer (REST) then aggregates these fused local subgraphs by explicitly injecting degree and physical distance priors, overcoming the over-smoothing limitations of traditional GNNs. Extensive experiments on real-world datasets for check-in, crime, service call and population prediction show that RAMEN consistently outperforms state-of-the-art baselines, achieving up to 35.3% MAE improvement. Visualizations of gating weights further suggest that RAMEN’s fusion aligns well with real-world urban physical characteristics. Full article

Task decomposition is a fundamental challenge in multi-agent collaborative maintenance systems, where unstructured natural language instructions must be precisely translated into logically coherent, executable sub-task sequences. This paper formulates task decomposition as a constrained optimal path search problem on a heterogeneous knowledge graph that encodes coal mine equipment topology, fault causality, and maintenance procedures. We construct a composite cost function that systematically integrates semantic similarity from graph neural network embeddings, relation-type weights, and structural path length, transforming instruction parsing into a mathematically tractable combinatorial optimization. The cost function is derived from the principles of shortest-path reasoning in knowledge graphs: the relational weights capture domain-specific association strengths, the semantic similarity term promotes contextually coherent chains, and the path-length penalty prevents unnecessarily long derivations. A multi-hop reasoning algorithm coupling heterogeneous graph convolution with beam search is developed to solve this problem efficiently, achieving high-quality approximate solutions while ensuring computational tractability. The reasoning process is inherently interpretable, as the optimal path directly maps to a traceable atomic task sequence with explicit dependency relations. A formal complexity analysis shows the algorithm scales as $O$($b$·$K$·$d_{max}$), where b is the beam width and d_max is the maximum node degree. Several theoretical properties of the proposed framework are further derived: the cost function is non-negative and strictly monotonic, optimal paths satisfy the optimal substructure property, cycle-free optimal paths always exist, and beam search can yield globally optimal solutions given a sufficiently large beam width. These theoretical conclusions establish mathematical guarantees for the presented decomposition framework. Experiments on 200 composite maintenance instructions with gold-standard annotations (inter-annotator agreement Cohen’s $\kappa =0.88$) demonstrate that the proposed method achieves 94.3% task sequence accuracy (95% CI: 91.2–96.8%) and 96.4% dependency accuracy (95% CI: 93.5–98.1%), substantially outperforming both a rule-augmented baseline (58.6%, 62.1%) and a GPT-4o few-shot chain-of-thought baseline (73.2%, 70.5%); McNemar’s test yields p < 0.001 for both comparisons. The average inference time is 29.7 ms (SD 2.1 ms), meeting stringent industrial real-time constraints. Ablation studies quantify the contribution of each cost function component and confirm the robustness of the chosen beam width and hyperparameters. When integrated into a full multi-agent system, the framework delivers end-to-end response time within 3 s (P50: 1.87 s, P95: 2.83 s) and maintains an 86.7% task success rate even under dual agent failures, validating the robustness of the proposed mathematical formulation. This work establishes a rigorous graph-theoretic foundation for instruction decomposition in multi-agent systems, with direct applicability to safety-critical industrial environments. Full article

Transient extrusion defects are a fundamental process control challenge in fused granular fabrication (FGF), arising from the finite melt pressurization time of screw-driven extruders at each start–stop event. This study presents an integrated experimental and analytical framework for the characterization and open-loop feedforward compensation of these transient dynamics, demonstrated on two composite thermoplastic feedstocks: a recycled random polypropylene (RPP1) and a 20 wt% wood-fiber-reinforced polypropylene composite (WFPP). Unlike prior filament- or single-material feedforward strategies, this framework derives and statistically validates material-specific compensation parameters across two rheologically distinct feedstocks. Single-layer road experiments were conducted on a custom instrumented FGF platform across a 2³⁻¹ half-fraction factorial design varying melt temperature, print acceleration, and nozzle diameter, with screw speed stepped among 20, 40, and 80 RPM to excite transient states; deposited road geometry was digitized and spatially registered to the synchronized process signals. Main-effects regression confirmed that nozzle diameter is the dominant predictor of mean road width, while screw velocity exerts a significant negative effect attributable to speed-dependent backflow. Prediction-error minimization on the pooled multi-experiment dataset yielded a parsimonious first-order transfer function, G(s) = 0.990/(1 + 1.909s), whose time constant is physically attributed to melt compressibility in the barrel volume upstream of the nozzle restriction. This model was embedded in a G-code post-processor implementing two sequential corrections: a material-specific steady-state slip gain and a discrete linear-advance term parameterized by the identified time constant. For RPP1 at the nominal gain, the print latency interquartile range decreased from 5–20 mm to 2–8 mm without degrading steady-state dimensional accuracy; the combination of nominal gain with active retraction further reduced latency to near-zero. Analysis of covariance (ANCOVA) confirmed that optimal feedforward gains are statistically material-dependent across all three quality metrics (p < 0.05), providing statistical justification for material-specific compensator parameterization. The results establish a practical, hardware-agnostic route to reduce transient deposition defects in pellet-based additive manufacturing, extensible to additional feedstocks. Full article

Landslide susceptibility mapping (LSM) serves as a fundamental technical support for geohazard prevention and mitigation across mountainous terrains. This research constructs a multi-scale terrain unit integrated modeling framework targeting complex mountainous geomorphic settings, taking Zhenping County as the research object. Multi-resolution digital elevation model (DEM) datasets, multi-source satellite remote sensing imagery (GF-2), geological vector datasets and hydrological survey data are jointly adopted as the basic data source. The r.slopeunits module embedded in GRASS GIS is utilized to automatically segment slope units, and a comprehensive composite index S, coupling slope partition quality indicator F and model prediction accuracy metric R, is proposed to adaptively optimize two critical slope-unit hyperparameters: circular variance (c) and minimum unit area (a). Four DEM spatial resolutions (15 m, 25 m, 50 m, 100 m) are systematically calibrated with 42 groups of c–a parameter combinations to screen out the optimal slope-unit segmentation scheme (c = 0.1, a = 200,000 m²). Twelve landslide predisposing covariates covering topography, hydrology, lithology, human engineering activities and land cover are selected after multicollinearity diagnosis via Variance Inflation Factor and mean utility factor contribution evaluation. Logistic regression tree (LMT), LMT-Adaboost and LMT-Random Subspace are compared by random cross-validation and spatial block cross-validation. Parameter sensitivity analysis is further carried out to quantify the stability of model outputs against DEM resolution and slope-unit parameter perturbations. The LMT-RSM ensemble achieved the highest spatial cross-validation AUC (0.954 ± 0.019), outperforming LMT (0.925 ± 0.023) and AdaBoost-LMT (0.934 ± 0.021). The DeLong test confirmed that LMT-RSM’s superiority over LMT is statistically significant (p < 0.0001). The proportion of landslides in the very high and high susceptibility zones under the LMT-RSM model reached 95.98%, demonstrating relatively excellent spatial discrimination. This study provides an operational framework combining optimized slope units, ensemble learning, and spatially explicit validation for robust LSM in complex terrain, and offers a reproducible technical pathway for landslide risk prevention in mountainous regions. Full article

Robust image-to-image correspondence is a fundamental challenge for camera-based Visual Simultaneous Localization and Mapping (SLAM). Conventional approaches primarily rely on isolated local feature matching or optical flow prediction, which often suffer from limited robustness under large parallax, high computational overhead, and cumulative drift errors during long-sequence tracking. To address these limitations, we propose SPTNet (SuperPoint Tracking Network), an efficient multi-task neural network that tightly couples feature detection, description, and dense optical flow prediction within a unified architecture. The fundamental innovation of SPTNet is a Hybrid Tracking Module (HTM) governed by a novel Predictor-Corrector mechanism. Specifically, the dense optical flow field acts as a temporal prior to constrain the descriptor matching search space, while the descriptors act as a correction signal, eliminating flow-induced drift at each frame through spatially constrained Sinkhorn optimization. This synergy enables efficient feature reuse via a shared backbone, minimizing redundant computation. Comprehensive experiments on indoor and outdoor datasets demonstrate that SPTNet attains a false matching rate as low as 1.8% at a 5-pixel threshold on HPatches, substantially reduces cumulative drift on long-sequence SLAM benchmarks, and maintains a high execution speed of 35 FPS on standard GPUs, demonstrating a highly compact footprint advantageous for prospective embedded robotic deployment. Full article

Rapid urbanization presents a fundamental challenge to global sustainable development, as urban expansion increasingly conflicts with resource constraints, ecological carrying capacity, and carbon emission mandates. Serving as a highly spatially heterogeneous laboratory, China offers a critical context for embedding green development into urban efficiency assessments. This study reconceptualizes new urbanization efficiency (NUE) through a multidimensional framework encompassing resource inputs, coordinated development processes, and sustainable outcomes. Using a panel of 281 prefecture-level cities, we evaluated NUE via a remote-sensing-constrained Super-SBM model, utilizing LISA time paths and an interpretable spatial machine learning framework (GWR-XGBoost-SHAP) to unpack its spatiotemporal dynamics. Key findings indicate: (1) China’s NUE exhibited a fluctuating upward trajectory, transitioning from an east-high/west-low to a south-high/north-low spatial pattern. (2) Spatiotemporal analysis revealed strong spatial inertia and profound path dependency in North China, characterizing it as a persistent low-value basin, whereas southeastern coastal cities demonstrated dynamic, path-breaking trajectories. (3) While green development intensity, industrial upgrading, and technological innovation emerged as primary drivers, they operate through complex nonlinear mechanisms. Specifically, green development and innovation exhibit threshold-triggered synergies, population agglomeration acts as a nonlinear amplifier, and external openness presents context-dependent negative interactions. These findings refine NUE measurement methodologies and provide a transferable analytical framework to inform differentiated, place-based urbanization policies for regions navigating the friction between urban growth and ecological limits. Full article

Virtual testing is one of the key methods for evaluating the collision avoidance capabilities of autonomous ships, and test scenarios provide the fundamental basis for its implementation. However, generating test scenarios that are both diverse and physically realistic remains a significant challenge in this field. This paper proposes a ship encounter scenario generation and dynamic simulation method for encounter situations based on a Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP). Specifically, the proposed framework consists of a Collision Risk Index-guided-WGAN-GP (CRI-WGAN-GP) for initial encounter scenario generation and a Physics-Informed Conditional WGAN-GP (PI-CWGAN-GP) for dynamic encounter scenario simulation. In the first stage, the CRI-WGAN-GP is developed to generate initial encounter scenarios with collision-risk characteristics. The Collision Risk Index (CRI) is directly embedded into the discriminator of the WGAN-GP. This mechanism forces the latent space to learn risk correlations, enabling effective exploration of high-risk boundaries and generating test cases with collision risks for the target ship. In the second stage, the PI-CWGAN-GP is designed to generate sequential dynamic simulation scenarios conditioned on the current encounter situation. By incorporating constraints related to ship motion performance, the PI-CWGAN-GP ensures that the generated ship behaviors are physically realistic. The proposed method is experimentally evaluated based on scenario diversity and physical realism, and the experimental results demonstrate that the proposed method can generate initial scenarios with collision risks and simulate ship behavior in encounter situations. Compared to existing random sampling and general generative adversarial network methods, the proposed approach shows significant advantages in generating more realistic and high-risk scenarios. In addition, a stress-test analysis is conducted based on a real encounter scenario, with the historical own-ship trajectory replayed as a reference response. Since no specific collision-avoidance algorithm is exposed to the generated scenarios in closed loop, the evaluation is limited to examining the encounter pressure imposed by the generated target-ship behaviors. The results show that the generated counterpart scenarios reduce the original spatial and temporal safety margins and create more demanding encounter conditions. Full article

Food safety culture (FSC) is a structured, measurable and enforceable element of the food industry, which is essential for building consumer trust and safeguarding consumer wellness. Regardless of the size of the organisation, FSC plays a pivotal role in safety and quality assurance and should be embedded in the company’s values and beliefs, ultimately manifesting in employee behaviours. FSC’s principal denominators encompass the fundamental principles of leadership, knowledge, engagement, environment, performance, and outcome. These principles collectively form a holistic framework that is pivotal for enhancing the efficacy and efficiency of safety initiatives in the food supply chain. FSC should be integrated into existing food safety standards, aligning with management systems and enjoying support and ownership at all levels and portfolios. Attaining a strong FSC requires commitment and active participation, not only from departments and sections directly involved with food handling, but also from administrative departments and branches. A non-conducive or weak culture, on the other hand, creates barriers to the achievement of safety goals and creates environments that may lead to product safety failures and non-conformances, with potential detrimental impacts on both consumer and business well-being. The absence of a specific culture constitutes a culture in itself, and, therefore, regular, valid, and reliable assessment is crucial for understanding the current state of FSC, without drawing generalised, superficial, or biased conclusions. In this study, the history and context of FSC are discussed, as well as narratives on assessment protocols and improvement initiatives. A trustworthy and ethical assessment is the first step in a three-phase process, involving assessment, alignment, and intervention to improve FSC, constituting various subcategories. The ultimate intent of this study is to provide turn-key solutions through presenting ideas, debating concepts and proposing interventions to guide and inform FSC improvement, culminating in safe and wholesome products. Full article

Geometry is typically treated as a primitive background structure within which physical processes unfold. In this work, we investigate an alternative possibility: that geometry can arise as a structural consequence of relational dynamics. We study the framework of Graph Dynamical Geometrization (GDG), in which a system is specified by a relational structure encoded by a graph, together with a dynamical law acting on that structure. The dynamics define a symmetric operator whose spectral evolution generates a positive–definite kernel. From this kernel, one obtains a squared Euclidean distance matrix that induces a canonical embedding of the relational system into a Euclidean space. Therefore geometry appears not as an assumed background but as a structure generated by the dynamics of relations. This construction motivates a philosophical interpretation that we call relational–dynamical structural realism, according to which the fundamental ontology consists of entities connected by relations and governed by a dynamical law, while geometric structure emerges from the spectral organization of those dynamics. We analyze the status of the induced geometry within several forms of structural realism. The GDG framework shows that invariant metric relations generated by relational dynamics can possess explanatory and predictive significance. More generally, the results suggest a structural principle: geometry may emerge whenever relational dynamics admit a spectral articulation. Full article

Artificial intelligence (AI) is increasingly embedded in high-stakes socio-technical systems, intensifying concerns about autonomy, accountability, data rights, and fundamental-rights protection. This article develops an exploratory, expert-informed Human-Centred AI (HCAI) pre-design governance framework that translates selected risk-based obligations of the EU Artificial Intelligence Act into early organisational decisions about human oversight, data accountability, documentation, and bounded algorithmic autonomy. Using a sequential mixed-methods design, the study combines an Analytic Hierarchy Process (AHP) survey of 28 experts with think-aloud interviews with 15 of those respondents. The AHP results show that, among the governance criteria included in the model, AI design objectives received the highest upper-level priority and human oversight and control received the highest global priority, followed by personal information protection, design ethics, intellectual property rights protection, and limits of algorithmic autonomy. The interviews explain these priorities by showing that experts framed trustworthy AI governance as a problem of controllability, responsibility allocation, traceable data use, rights protection, and verifiable human intervention rather than model performance alone. The study contributes by defining pre-design governance as a bounded initial consideration-stage decision structure, combining AHP-based priority evidence with qualitative justification logic, and proposing a preliminary governance package of decision points, minimum evidence artefacts, and illustrative operational check criteria. The package is not presented as a validated legal compliance model; instead, it provides an expert-informed translation pathway for future organisational, sector-specific, and empirical validation. Full article

In edge computing scenarios for remote sensing image interpretation, two fundamental challenges constrain the effectiveness of federated fine-tuning: the limited computational capacity of edge devices restricts the multi-scale feature learning capability of lightweight deployed models, while the highly heterogeneous and imbalanced data distributions (Non-IID settings) across clients render conventional parameter aggregation strategies ineffective. To address these challenges jointly, this paper proposes Fed-RSAdapter, a federated fine-tuning framework for remote sensing imagery based on multi-scale adapter modules. On the edge side, a multi-scale adapter architecture with cross-layer dense connections and feature map concatenation alignment is introduced. By inserting lightweight adapter modules in parallel within a frozen backbone, the proposed design captures multi-scale spatial characteristics inherent in high-resolution remote sensing imagery while confining trainable parameters to less than 5% of the full model, thereby satisfying strict on-device computational and communication constraints. On the server side, a parameter similarity-aware aggregation strategy is designed to handle client heterogeneity. Client adapter parameters are normalized and mapped into a similarity matrix via Gaussian kernel distance, enabling the server to perform personalized weighted aggregation that balances global generalization with client-specific adaptation. Extensive experiments on scene classification, object detection, and semantic segmentation benchmarks demonstrate that Fed-RSAdapter achieves an average improvement of 3.75% in overall accuracy for scene classification, 0.74% in mAP for object detection, and 0.25% in overall accuracy for semantic segmentation over federated baselines, while reducing the volume of transmitted parameters to approximately 4.93% of that required by conventional full-parameter federated learning. These results demonstrate the communication-efficient fine-tuning capability of the proposed framework under federated remote sensing settings. We further clarify that actual deployment on embedded edge hardware is not directly evaluated in this work and is therefore treated as an important direction for future validation rather than an experimentally proven conclusion. Full article

Vascularization remains a fundamental bottleneck in tissue engineering, as the absence of functional vascular networks limits oxygen and nutrient delivery, resulting in necrotic cores and poor host integration. While structural scaffold design and cell sourcing have advanced considerably, emerging evidence indicates that paracrine signaling, rather than direct cell contact or scaffold architecture alone, is the primary driver of angiogenesis and vasculogenesis within engineered constructs. Key cell types, including endothelial cells (ECs) and mesenchymal stem cells (MSCs), engage in bidirectional paracrine crosstalk through the secretion of vascular endothelial growth factor (VEGF), angiopoietins, hepatocyte growth factor, and platelet-derived growth factor, among other mediators. While researchers have long focused on improving scaffold structure and cell selection, growing evidence shows that the chemical messages cells send to one another play a far more important role in driving blood vessel formation than previously appreciated. This review explores how cells embedded within engineered scaffolds communicate through secreted signals to coordinate the growth of new blood vessels. Two cell types, MSCs and ECs, are central to this process: cells that line blood vessels and bone marrow-derived stem cells. These cells exchange a variety of chemical messages that instruct neighboring cells to multiply, move, and organize into vessel-like structures. Importantly, the material properties of the scaffold itself, including its stiffness, surface texture, and degradation over time, influence the signals cells produce and how those signals spread through the tissue. Strategies to amplify paracrine signaling include growth factor-loaded delivery systems, hypoxic and genetic preconditioning of MSCs, and perfusion bioreactor culture. In vitro and in vivo evidence consistently demonstrates that coculture systems leveraging paracrine interactions produce superior vascular outcomes compared to single-cell or acellular constructs. Despite this progress, challenges related to signaling complexity, reproducibility, and clinical translation persist. Integration of transcriptomic and proteomic profiling, computational modeling, and machine learning approaches offers a path toward rationally designed scaffolds that recapitulate the spatiotemporal dynamics of native vascular signaling and ultimately support functional tissue regeneration. Full article

Inverse kinematics (IK) is fundamental to robot motion planning. Classical analytical solvers require complete Denavit–Hartenberg (DH) parameters that are often proprietary or degraded by mechanical wear, and numerical solvers based on damped least squares (DLS) are sensitive to initialization, particularly near singularities. We propose XGNN, an explainable graph neural network positioned as a model-free, interpretable warm-start initializer for downstream numerical IK refinement rather than as a standalone replacement for analytical solvers. Each IK query is encoded as a 12-node graph in which six pose nodes and six joint nodes are connected through bipartite pose-to-joint attention edges and chain edges along the kinematic structure. GATv2 message passing aggregates information at each joint node; two ablation-validated design contributions (a learnable node-type embedding and an angle-aware composite loss) enable training to convergence. Evaluated on 300,000 trajectory-style samples generated from the ABB IRB 2400 kinematic model, XGNN achieves ${3.66}^{\circ }$ joint mean absolute error (MAE), comparable to a multilayer perceptron baseline (${3.09}^{\circ }$) and a bidirectional LSTM (${3.14}^{\circ }$) under identical training. The standalone joint accuracy of all learned models is too coarse for direct industrial use, but XGNN provides the strongest warm start for DLS refinement: the convergence rate improves from $98.4\%$ to $100\%$, mean iterations drop from 14.6 to 3.2, and wall-clock time per pose drops $5.0\times$ on the IRB 2400. The benefit transfers cross-platform to the Universal Robots UR5 collaborative manipulator (convergence rate $82.2\%$ to $100\%$, $10.0\times$ speedup) and survives DH parameter perturbation of up to $\pm 10\%$, simulating calibration drift or mechanical wear. The GATv2 attention coefficients additionally provide an interpretability signal at zero inference cost. XGNN therefore complements analytical and numerical IK methods as an interpretable, calibration-robust warm start when DH parameters are unavailable, proprietary, or degraded. Full article

Two-dimensional (2D) phase unwrapping (PU) in interferometric synthetic aperture radar (InSAR) remains a challenging ill-posed inverse problem, particularly in low-coherence regions where the Itoh condition is violated. Although deep-learning methods have shown promise, existing networks often struggle to preserve phase continuity and structural details under strong decorrelation noise. To address these limitations, a novel physics-guided conditional generative adversarial network (cGAN) is proposed in this work. Specifically, an adaptive atrous spatial pyramid pooling module mimics the multi-baseline observation mechanism to resolve multi-scale phase variations. Meanwhile, a multi-channel attention-augmented discriminator guides adversarial learning to focus on error-prone low-coherence regions. Additionally, a hybrid loss function enforces fundamental physical constraints, including phase-wrapping consistency and path continuity, explicitly embedding the prior knowledge of interferometric phase physics into the data-driven framework. Extensive experiments demonstrate that the proposed method achieves highly competitive performance in terms of root mean square error (RMSE) and structural similarity (SSIM) compared to State-of-the-Art approaches, especially in scenarios with severe noise and discontinuities. Full article