Search Results (331)

In recent years, reinforcement learning (RL) has been applied to frontier-based exploration to enhance a robot’s decision-making policy and improve exploration performance. In this work, we address this scenario with the aim of pushing forward the finding of the optimal frontier selection policy in unknown, unstructured environments, with RL deployed for a minimal sensing drone setup. We propose a novel policy architecture, featuring an attention module that uses the global map features captured by a convolutional neural network together with local frontier features in the form of scalar values, trained end-to-end with a scoring network using the Proximal Policy Optimization algorithm over a 2D randomized unstructured environment. Our approach demonstrates improved exploration efficiency in the evaluated scenarios, as it surpasses purely heuristic-based frontier selection strategies used as baselines for other RL methods, achieving shorter paths than the Nearest Frontier, the Hybrid Approach, and the TARE local horizon, as well as one-shot sim-to-real policy deployment. Full article

As artificial intelligence (AI) becomes increasingly important in rural revitalization, building consensus among multiple stakeholders and developing participatory digital co-creation platforms has grown increasingly urgent. However, existing large language model (LLM) systems predominantly adopt a one-shot generation paradigm, making it challenging to accurately capture villagers’ cultural aspirations and frequently resulting in a significant disconnect between design outputs and community expectations. This situation reveals deficiencies in progressive deliberation mechanisms and cultural controllability. To address these issues, this study proposes a multimodal Participatory Landscape Demand Generation (PLDG) system to enhance AI-generated dialogue controllability, facilitate effective cultural translation in sensitive rural contexts, and promote sustainable development where landscape design both drives and reflects rural revitalization. The system leverages LLMs to simulate stakeholder participatory interactions in village landscape design scenarios. Using culturally distinctive Chinese villages as case studies, the research conducts multi-role simulated dialogues, multimodal semantic extraction, and iterative consensus-building, and evaluates the resultant data to generate landscape design proposals. The results indicate that the PLDG system significantly improves participation efficiency among diverse design stakeholders and enhances the sustainability of design decisions. Compared to conventional methods, metrics such as cultural compatibility, villager participation, and design innovation show substantial improvements. These findings demonstrate the considerable potential of human-AI collaboration in future rural planning. This study introduces the Culture Constraint-Driven Rural Landscape AI Collaborative Design Framework (PLDG), validating its practical efficacy in identifying culturally sensitive elements, ensuring cultural congruence, facilitating community participation, and fostering design innovation. Consequently, it provides a reusable, iterative operational tool for the digital renewal of sustainable rural landscapes. Full article

Binary Waring decomposition seeks to express a homogeneous binary form as a minimal sum of powers of linear forms. In the binary setting, Sylvester’s theorem gives a classical algebraic route for rank determination and parameter recovery through structured Hankel/catalecticant matrices. Although this procedure is exact and interpretable in ideal arithmetic, practical rank identification may become unstable when the input coefficients are contaminated by noise or when the underlying roots are close to degenerate configurations. This paper develops a data-driven rank inference framework coupled with certified Sylvester reconstruction for robust binary Waring decomposition. The proposed method first converts the coefficient sequence into a Hankel-aware graph that captures recurrence-induced dependencies among polynomial coefficients. A graph neural network is then used to infer plausible rank candidates from this structured representation. Instead of accepting a single prediction directly, the framework performs explicit Sylvester reconstruction and algebraic residual verification for candidate ranks. To further improve decision reliability, a lightweight meta-verification module integrates reconstruction residuals, model confidence scores, and stability-related indicators to select the most credible rank. Experiments on large-scale synthetic binary forms show that the proposed meta-guided variant improves rank identification and verified reconstruction success relative to the one-shot hybrid solver under low-to-moderate noise while maintaining the transparency and auditability of classical symbolic–numeric computation. Additional stress tests indicate that performance can degrade under shifted sampling regimes; so, the method should be interpreted as a robust decision layer within the modeled problem class rather than as unconstrained real-world validation. Full article

Automating system-level nuclear thermal-hydraulic (T-H) model construction remains challenging because platform-specific API syntax, graph connectivity, parameter dependency ordering, and solver admissibility must be satisfied simultaneously. This study develops a closed-loop modeling framework on the SAFRI platform by combining supervised fine-tuning (SFT), a Plan-and-Act agent with retrieval-grounded parameter completion, and reinforcement learning based on group relative policy optimization (GRPO). The SFT stage uses a 6003-record domain corpus derived from expert-authored or expert-verified SAFRI modeling exemplars, while system-level generalization is evaluated on a held-out 50-case in-house evaluation set separated at the case-template level. At the component level, LoRA-adapted Qwen3-8B achieves 100% code accuracy, compared with 50% for zero-shot and 74% for one-shot prompting. At the system level, the SFT agent attains a 100% syntax success rate (SSR), 90% topology success rate (TSR), and 72.4% physical convergence rate (PCR), showing that local API correctness is insufficient for solver-valid model assembly. After GRPO training with schema, topology, physics, and sequence rewards, the full SAFRI-SFT-RL agent reaches a 100% SSR, 100% TSR, and 88.8% PCR on the in-house evaluation set, while an error self-healing loop resolves execution-time failures in an average of 2.3 corrective iterations. These results show that solver-grounded reinforcement learning is effective for closing the gap between syntactically correct script generation and physically convergent nuclear T-H model construction. Full article

Robotic sorting-and-filling tasks in flexible manufacturing require robots to reproduce specified in-box arrangements while adapting to variations in container poses, object availability, sensing conditions, and external interventions. This paper proposes a box-centric one-shot teaching framework for robotic packing tasks with relative pose constraints. In the teaching stage, a human operator demonstrates the desired packing layout only once. The system uses reference-prompted SAM-based contour refinement to extract box and in-box object contours, object categories, quantities, and relative position and orientation constraints. These constraints are then converted from pixel-plane measurements into box-local pose constraints, forming a reusable box-centric packing template that preserves both translational and angular layout information. During execution, the recorded template is transferred to detected box instances with different global poses, and executable pick-and-place commands are generated through a task-level perception-to-command pipeline. A mechanism for continuous assignment and state updates is further introduced to maintain residual target slots, update object-to-slot allocation, and report missing or redundant objects across execution rounds. Single-box template transfer experiments achieved mean placement errors of 7.16 mm and 7.57 mm for two recorded templates, while representative post-execution images further showed that the relative object orientations were visually preserved with respect to the taught template footprints. Multi-box experiments demonstrated that unfinished residual slots could be preserved and completed after scene updates without re-teaching. Additional validation with different container types and object shapes showed the feasibility of extending the framework beyond cube-only cases. Ablation tests under nine exposure settings further showed that SAM refinement improved template-acquisition robustness compared with the previous recognition method. These results verify that the proposed framework enables one-shot template acquisition, box-centric layout transfer, relative pose preservation, and persistent task-level execution for constrained robotic packing tasks. Full article

Large language models (LLMs) are advancing intelligent writing systems from local text continuation and language polishing toward long-form structured text generation. However, directly generating full-length academic paper drafts remains challenging due to unclear research objectives, unstable discourse structures, insufficient long-text coherence, and the lack of explicit quality control mechanisms. To address this long-form structured generation task, we propose MetricDraft, a metric-driven framework for academic paper draft generation. The framework organizes the drafting process as a closed-loop pipeline comprising research ideation clarification, structural anchoring, section-by-section generation, quality assessment, and feedback-driven revision. Its key components include adversarial research ideation clarification, staged structural anchoring, the PRISM structured metric system, progressive context injection with section-type-aware guided generation (PCI+STAGG), and a metric-feedback-driven generation–evaluation co-optimization mechanism. Experimental results demonstrate that MetricDraft achieves higher composite quality scores than one-shot generation, summary-based context passing, and context-accumulation-only baselines, improving MQS over Base1, Base2, and Base3 by +5.5, +7.9, and +7.0 points, respectively, with paired tests reaching statistical significance. To examine whether this advantage is tied to a single LLM backend, we further conduct a cross-model validation on all 15 tasks using Qwen3.7-Max in addition to the original DeepSeek-V4-Pro setting. MetricDraft remains the best-performing strategy under both models. To address citation reliability, an additional citation verification-and-retrieval-based replacement (CVRR) experiment reduces the fabricated citation rate of DeepSeek MetricDraft drafts from 56.0% to 15.0%. Furthermore, PRISM exhibits moderate-to-high positive correlations with expert ratings, providing preliminary evidence that it can serve as an auxiliary evaluation reference for draft quality diagnosis and iterative revision. This work reformulates academic writing as an adjustable, assessable, and iteratively optimizable long-form structured text generation problem, offering methodological insights for human–AI collaborative writing and intelligent text generation system design. Full article

Plant disease recognition models trained under laboratory conditions often degrade markedly after cross-domain transfer because of the pronounced distribution gap between source and target domains and the scarcity of labeled target-domain samples. To address the transfer task from PlantVillage (PV_100) to PlantDoc, this study develops and evaluates a lightweight cross-domain few-shot plant disease recognition method under a strict PlantVillage-to-PlantDoc protocol. The method integrates EfficientNet-B0 feature extraction, cosine-similarity-based prototypical classification, and training-time target-domain BN adaptation (TBA). During training, unlabeled target-domain images are used only for BN statistical calibration, whereas inference is limited to feature extraction and prototype matching, without gradient updates or iterative optimization. Under a unified experimental protocol, the proposed method achieved cross-split mean accuracies of 42.69 ± 0.62% for one-shot and 54.24 ± 0.72% for five-shot, where ± denotes the standard deviation across three strict data splits; it outperformed ProtoNet by 7.44 and 9.43 percentage points, respectively. Ablation results indicate that TBA is the main source of performance improvement, whereas more complex adaptation strategies do not yield stable additional gains. The core encoder can be executed entirely on the NPU, with an estimated single-sample inference latency as low as 0.658 ms, indicating strong potential for encoder-level mobile deployment. Full article

Complex agricultural question answering often requires multi-step reasoning over domain-specific knowledge and reliable retrieval of heterogeneous evidence. However, existing retrieval-augmented generation (RAG) methods usually rely on one-shot retrieval and provide limited control over whether the retrieved evidence is sufficient, accurate, and consistent for answering complex agricultural questions. To address this limitation, this paper proposes an agent-driven question-answering framework for the agricultural planting domain based on a Planning–Execution–Feedback (PEF) closed-loop mechanism. The framework decomposes complex queries into executable subtasks, performs knowledge acquisition through a knowledge-graph-guided hybrid retrieval module, and iteratively refines the reasoning process according to retrieval-quality feedback. Specifically, in the retrieval stage, a two-stage strategy is introduced to first localize candidate entities in the knowledge graph and then conduct context-enhanced dense retrieval with entity-consistency reranking, thereby reducing semantic drift and improving domain alignment. In the feedback stage, the agent evaluates the adequacy of the retrieved evidence and determines whether to continue execution, re-retrieve evidence, or replan the workflow. Experimental results on the AgroQA dataset show that the proposed method achieves 88.9%, 79.1%, and 92.6% on the Answer-C, Answer-R, and CR metrics, respectively, outperforming traditional retrieval-augmented and general large language model baselines. In addition, a three-dimensional digital human interface is implemented as an application prototype to demonstrate the feasibility of integrating the proposed framework into interactive agricultural knowledge services. Full article

Accurate anatomical landmark detection is important for orthodontic analysis, surgical planning, and morphometric measurement, but fully supervised methods usually require large expert-annotated datasets. This work studies a one-shot setting, where only a single annotated template image is used for training. We propose a foundation-model-based landmark detection framework using a frozen DINO Vision Transformer (ViT) backbone. The proposed framework integrates three complementary components: a Multi-Layer Multi-Facet (MLMF) module that adaptively fuses key and value features from multiple ViT layers through global source-wise reweighting; a Mamba-Based Long-Range Context Aggregation (MLCA) module that injects global anatomical context into fused patch descriptors with linear complexity; and a Topology-Constrained Graph Refinement (TCGR) module that refines the predicted landmark configuration using anatomical graph constraints. Experiments on the Cephalometric dataset and the Hand X-ray dataset demonstrate that the proposed method achieves strong performance. Overall, the results show that jointly exploiting multi-source foundation-model representations, efficient long-range context aggregation, and topology-aware refinement improves annotation-efficient anatomical landmark detection. Full article

One-shot devices yield binary current-status observations because each unit can be tested only once. In stockpile settings, where testing is destructive and failures are rare, fixed-sample plans may consume more units than necessary. This paper proposes a sequential estimation procedure for monitoring one-shot device stockpiles at a fixed inspection time. The method is formulated in terms the inspection-time failure probability and its reciprocal, and is therefore applicable to general one-parameter lifetime models. A squared-error-plus-sampling-cost risk function is used to derive a stopping rule that adaptively determines the required number of observed failures. We establish first-order efficiency relative to an optimal fixed-sample benchmark and illustrate the framework through a stockpile case study motivated by stored N95 respirators. The proposed procedure provides a practical and theoretically justified tool for stockpile monitoring when destructive testing is costly and unnecessary sampling should be minimized. Full article

This work presents, for the first time, a novel cross injection analysis (CIA) system for the simultaneous multi-determination of key biomarkers associated with diabetic nephropathy—namely, albumin, creatinine, and glucose—within a single analytical run. Unlike conventional flow-based techniques that rely on sequential measurements, the proposed CIA platform integrates multiple analytical pathways into a unified design, enabling one-shot multi-analyte analysis without the need for complex separation units or injection valves. The system employs peristaltic pumps and a rectangular platform with orthogonal flow channels, allowing concurrent aspiration and efficient transport of reaction products to compact detectors. Albumin determination was based on ion-association with tetrabromophenolphthalein ethyl ester. Creatinine was measured using the Jaffé reaction. Glucose was colorimetrically detected via its reaction with 3,5-dinitrosalicylic acid. The developed CIA provides enhanced sensitivity through its pre-mixing effect, enabling reliable quantification of trace analytes. Excellent analytical performance was achieved, including wide linear ranges (r² > 0.99), good precision (RSD < 7%), and rapid analysis (5 min). The method was validated against established reference methods, showing no significant differences, and successfully applied to urine with satisfactory recoveries (84.8–107.3%). Importantly, the proposed system adheres to green chemistry by minimizing reagent consumption and waste generation, offering a sustainable approach for multi-parameter clinical analysis. Full article

Remote sensing scene classification (RSSC) faces severe challenges from data scarcity and complex background clutter. To overcome these limitations, this paper draws inspiration from few-shot fine-grained image classification (FSFGIC) to filter noise and capture subtle details. However, existing methods often process global context and local features separately, which limits their ability to suppress background noise in complex scenes. Consequently, the Guided Relational Cross-Attention Dual-branch Network (GRCD-Net) is proposed. Its core Guided Relational Cross-Attention (GRC) block leverages global semantics to filter local background noise prior to bidirectional feature interaction. Additionally, Iterative Global Relation (IGR) and Patch-level Dual-Metric (PDM) modules are integrated to robustly refine global relations and capture local similarities. Extensive experiments demonstrate that GRCD-Net consistently outperforms baselines by 2–4% on standard FSFGIC benchmarks. Notably, on the challenging NWPU-RESISC45 RSSC dataset, it achieves an 81.39% one-shot accuracy and exceeds current state-of-the-art methods by 7.55%, validating its efficacy for complex Earth observation. Full article

Speech-driven 3D cartoon facial animation remains underexplored due to the difficulty of handling heterogeneous geometries with exaggerated proportions, limited generalization to diverse unseen subjects, and the scarcity of datasets. To address these challenges, we propose Let Toon Talk, a two-stage cascaded framework that effectively mitigates these bottlenecks in both modeling and driving. It enables one-shot, speech-synchronized 3D animation from a single unseen humanoid cartoon image, driven by arbitrary audio. Specifically, for avatar modeling, we propose a parametric adaptation mechanism to capture diverse heterogeneous facial topologies, which subsequently guides a feed-forward reconstruction module to create high-quality 3D Gaussian Splatting (3DGS) avatars. Building upon this, for speech driving, we introduce an Identity-Adaptive Flow Matching network. This generative module effectively maps audio to precise facial dynamics, achieving identity-adaptive motion synthesis for diverse humanoid cartoon characters without per-subject pretraining. Furthermore, we construct a hybrid cartoon talking-face dataset with a systematic curation strategy to bridge the data gap. Extensive experiments demonstrate that our framework produces high-quality, temporally coherent animations, exhibiting effective generalization on unseen structurally humanoid cartoon characters. Full article

Indoor localization for unmanned aerial vehicles (UAVs) remains challenging in GNSS-denied environments due to the difficulty of position calibration of multiple ultra-wideband (UWB) anchors and the asynchronous fusion of heterogeneous sensors. This paper proposes a multi-sensor fusion localization framework that integrates multi-anchor one-shot calibration with factor graph optimization (FGO). First, Landmark Multidimensional Scaling (LMDS) is used to reconstruct the relative geometry of the anchors and the onboard tag from ranging measurements. Then, rigid Procrustes alignment is performed using a small number of anchors with known coordinates in the East–North–Up (ENU) frame to recover the transformation to the ENU frame, thereby enabling efficient position calibration of multiple UWB anchors and UAV pose initialization. Subsequently, a tightly coupled factor graph is constructed by incorporating inertial measurement unit (IMU) pre-integration, UWB ranging, laser rangefinder height measurements, and visual–inertial odometry (VIO) pose constraints. The resulting nonlinear optimization problem is solved using incremental smoothing, which improves robustness against non-line-of-sight (NLOS) errors and long-term drift. Experimental results on anchor calibration, public datasets, and real-world indoor UAV flights demonstrate that the proposed method improves the accuracy and robustness of indoor UAV localization. In particular, on the real-world rectangle trajectory, FGO-TC reduces the RMSE by approximately 38.8% compared with FGO-LC. Full article

The combination of the dry fiber AFP preforming process and RTM injection process brings new possibilities with regard to automation, high-quality manufacturing, and high-performance characteristics for out-of-autoclave composite manufacturing, particularly in aerospace industry. This paper describes the manufacturing of an aircraft engine Outlet Guide Vane (OGV), made with a dry carbon fiber preform manufactured with Automated Fiber Placement (AFP) and co-injected, co-cured, and co-bonded with titanium fittings through the Resin Transfer Molding (RTM) Process. The details of the assembly process and necessary steps are described. Parts of the digitalization process behind the manufacturing are described, including information about integrated sensors and data management. Full article