Search Results (404)

Recent advances in large language models (LLMs) have made automated multiple-choice question (MCQ) generation increasingly feasible; however, reliably producing items that satisfy controlled cognitive demands remains a challenge. To address this gap, we introduce ReQUESTA, a hybrid, multi-agent framework for generating cognitively diverse MCQs that systematically target text-based, inferential, and main idea comprehension. ReQUESTA decomposes MCQ authoring into specialized subtasks and coordinates LLM-powered agents with rule-based components to support planning, controlled generation, iterative evaluation, and post-processing. We evaluated the framework in a large-scale reading comprehension study using academic expository texts, comparing ReQUESTA-generated MCQs with those produced by a single-pass GPT-5 zero-shot baseline. Psychometric analyses of learner responses assessed item difficulty and discrimination, while expert raters evaluated question quality across multiple dimensions, including topic relevance and distractor quality. Results showed that ReQUESTA-generated items were consistently more challenging, more discriminative, and more strongly aligned with overall reading comprehension performance. Expert evaluations further indicated stronger alignment with central concepts and superior distractor linguistic consistency and semantic plausibility, particularly for inferential questions. These findings demonstrate that hybrid, agentic orchestration can systematically improve the reliability and controllability of LLM-based generation, highlighting workflow design as a key lever for structured artifact generation beyond single-pass prompting. Full article

DUR (Deep Underground Railways) stations, such as Suseo Station in Korea, present unique evacuation challenges stemming from multi-level spatial depth, long vertical circulation paths, and rapid smoke spread dynamics. Conventional design guidelines often fail to capture these complexities, underscoring the need for advanced, simulation-driven safety evaluation frameworks. This study proposes a comprehensive Digital Twin-based methodology that integrates spatial topology modeling, agent-based evacuation simulation, and dynamic hazard-aware routing. A multi-layer map topology was constructed from high-fidelity architectural geometry, decomposing the station into functional regions and encoding connectivity across platforms, concourses, corridors, and vertical circulation elements. Real-time hazard conditions were reflected through dynamic adjustments to edge weights, allowing evacuation paths to adapt to blocked exits, fire shutter operations, and smoke-infiltrated domains. Ten evacuation scenarios were developed to assess sensitivity to fire origin, exit availability, vertical circulation failures, and onboard passenger loads. Simulation results reveal that evacuation performance is primarily constrained by vertical circulation bottlenecks, with emergency stairways (E1 and E2) serving as critical choke points under high-density conditions. Cases involving exit closures or fire-compartment failures produced significant delays, frequently exceeding NFPA 130 and KRCODE performance criteria. Conversely, guided evacuation strategies demonstrated marked improvements, reducing congestion and enabling compliance with platform evacuation thresholds even in full-load scenarios. These findings highlight the necessity of transitioning from static design evaluations toward Digital Twin-enabled, predictive safety management. The proposed framework enables real-time visualization, intervention testing, and operator decision support, offering a scalable foundation for next-generation evacuation planning in extreme-depth railway infrastructures. Full article

Advances in artificial intelligence (AI), particularly agentic AI, have created opportunities to enhance global sustainability by improving the efficiency and accuracy of environmental monitoring and response systems. Agentic AIs autonomously plan and execute towards specific goals with minimal or no human intervention; however, accessing environmental data is challenging and requires expertise due to inherent fragmentation and the diversity of data formats. The Model Context Protocol (MCP) is an open standard that allows AI systems to securely access and interact with diverse software tools and data sources through unified interfaces, reducing the need for custom integrations while enabling more accurate, context-aware assistance. This study introduces WeatherInfo_MCP, an interface that provides the required expertise for AI agents to access National Weather Service (NWS) data. Built on a service-oriented architecture, the system uses a centralized engine to handle robust geocoding and data extraction while providing AI agents with simple, independent tools to retrieve weather data from the NWS API. The system was validated through 14 unit tests and 23 comprehensive protocol compliance tests against the MCP 2025-06-18 specification, achieving a 100% pass rate across all categories, demonstrating its reliability when working with AI agents. We also successfully tested our model alongside a memory MCP to showcase its performance in a multi-MCP environment. While in its earliest version, WeatherInfo_MCP connects to the NWS API, its modular design and compliance with software development and MCP standards facilitate immediate expansion to additional environmental data and tools. WeatherInfo_MCP is released as an open-source tool to support the sustainable development community, enabling broad adoption of AI agents for environmental use cases. Full article

This paper introduces a multi-turn large language model (LLM) jailbreaking attack called Defamiliarization, in which malicious queries are embedded within ostensibly harmless narratives. By reframing requests in “unmarked” contexts, LLMs can be coerced into producing undesirable outputs. A range of scenarios is documented, from planning ethically dubious actions to selectively overlooking critical events in literary texts, thereby exposing the limitations of alignment strategies predicated on detecting trigger words or semantic cues. Rather than substituting vocabulary, defamiliarization manipulates context and presentation, highlighting vulnerabilities that cannot be addressed by token-level fixes alone. Beyond demonstrating the effectiveness of defamiliarization as an attack strategy, evidence is presented of a systematic relationship between model scale and susceptibility. Experiments reveal that smaller-parameter models are significantly easier to manipulate using defamiliarized prompts. This finding raises important concerns regarding the growing popularity of lightweight, locally hosted LLMs, which are favored for their lower computational requirements but may lack alignment safeguards. A more holistic approach to LLM safety is advocated—one that incorporates insights from literary theory, ethics, and user experience—treating these models as interpretive agents. By doing so, defenses against covert manipulations can be strengthened and AI systems can remain aligned with human values. Full article

Addressing the multi-constraint, nonlinear optimization challenge of trajectory planning for multi-rotor agents in urban high-rise environments, this paper proposes an improved hybridization of Harris hawks optimization (HHO) with a pigeon-inspired optimization (PIO) algorithm, termed improved hybridization of Harris hawks with pigeon-inspired optimization (IHHHPIO). Conventional intelligent optimization algorithms often suffer from slow convergence rates or susceptibility to local optima in such complex scenarios. This research establishes a hierarchical collaborative search framework, where the HHO algorithm acts as a top-level coordinator for global exploration and region allocation, while the PIO algorithm functions as a bottom-level searcher for fine-grained optimization within designated areas. The two algorithms collaborate through a bidirectional information exchange mechanism: HHO guides the local search direction of each PIO group with global best-position information, and each PIO group feeds back its locally optimal solutions to HHO for updating the global optimum. Simulation results demonstrate that the proposed IHHHPIO algorithm significantly outperforms both standard PIO and HHO algorithms in terms of convergence speed, solution accuracy, and stability, effectively planning safe, efficient, and collision-free flight trajectories. This work provides a reliable solution for agent logistics applications in complex urban environments. A certain limitation of this work lies in its validation solely through simulation, without physical experimental verification. Full article

Propelled by the “dual-carbon” strategy, green and intelligent ports are rapidly advancing toward low-carbon and intelligent development. However, the large-scale incorporation of renewable energy and the extensive electrification of transport equipment have substantially heightened system volatility and scheduling complexity. To address the challenges associated with multi-energy coupling and economic operation in medium and large ports, a hierarchical collaborative optimization scheduling strategy is proposed. The upper layer employs an improved Bio-enhanced Dung Beetle Optimization (BDBO) algorithm for parameter optimization and carbon-cost minimization. Meanwhile, the lower layer establishes a rolling time-series control mechanism grounded in Adaptive Dynamic Hierarchical Decoupling Planning (ADHDP), thereby constituting an integrated BDBO-ADHDP dual-agent system. Simulation results across four seasonal scenarios demonstrate that the proposed methodology outperforms DQN, PSO, GA, ACO, and DBO algorithms in reducing grid power purchases, enhancing renewable energy utilization, mitigating curtailment, and lowering operational costs. Moreover, it achieves faster convergence, superior robustness, and effective carbon-emission control. This study substantiates the efficacy of the proposed strategy within green port integrated energy systems and highlights its potential for broader application in other multi-energy coupled systems. Full article

Deep-sea exploration increasingly relies on Autonomous Underwater Vehicles (AUVs) to enable persistent, wide-area surveying in harsh and uncertain environments. In practice, however, deployments are constrained by tight energy budgets and bandwidth-limited, intermittent acoustic links, which complicate mission-level coordination. Moreover, many existing systems treat perception and control as loosely coupled modules, often resulting in redundant sensing, inefficient communication, and degraded overall performance—particularly under heterogeneous sensing modalities and shifting geological conditions. To address these challenges, we propose a hierarchical Federated Meta-Transfer Learning (FMTL) framework that tightly integrates collaborative perception with adaptive control for swarm optimization. The framework operates at three levels: (1) Representation Learning aligns heterogeneous sensors in a shared latent space via a physics-informed contrastive objective, substantially reducing communication overhead; (2) Meta-Learning Adaptation enables rapid transfer and convergence in new environments with minimal data exchange; and (3) Energy-Aware Control realizes closed-loop exploration by coupling Federated Explainable AI (FXAI) with decentralized multi-agent reinforcement learning (MARL) for path planning under energy constraints. Validated in high-fidelity hardware-in-the-loop simulations and a digital-twin environment, FMTL outperforms state-of-the-art baselines, achieving an AUC of 0.94 for target identification. Furthermore, an energy–intelligence Pareto analysis demonstrates a 4.5× improvement in information gain per Joule. Overall, this work provides a physically consistent and communication-efficient blueprint for the optimization and control of next-generation intelligent marine swarms. Full article

Maritime dynamic observation missions, such as environmental monitoring, marine ranching inspection, and emergency response, typically require large-scale and high-efficiency operations in complex and variable maritime environments. Unmanned Surface Vehicles (USVs) and Unmanned Aerial Vehicles (UAVs) offer complementary advantages in such missions: USVs provide long endurance and stable platform support, while UAVs enable rapid, high-coverage aerial perception. However, limited UAV battery capacity and dynamic task environments pose significant challenges to autonomous collaborative operations. This study proposes a collaborative operation and energy replenishment strategy for USV–UAV systems in maritime dynamic observation missions. Under a unified framework, task allocation, collaborative path planning, and energy replenishment are jointly optimized, where the USV serves as a mobile replenishment platform to provide energy support for the UAV. The proposed method incorporates dynamic task updates, environmental disturbances, and energy constraints, achieving real-time adaptive collaboration between heterogeneous agents. Validation through both simulations and actual sea trials demonstrates that the proposed strategy significantly outperforms four baseline methods (greedy strategy, static planning, multi-objective genetic algorithm, and reinforcement learning scheduler) across five core metrics: task completion rate (91.74% in simulation/90.85% in sea trials), total energy consumption (1284.66 kJ/1298.42 kJ), mission completion time (40.28 min/41.12 min), average response time (10.21 s/10.35 s), and path redundancy (13.79%/14.03%). Furthermore, ablation experiments verify that the energy replenishment strategy enhances the task completion rate in both simulation and field tests. This method provides a feasible and scalable collaborative solution for autonomous multi-agent systems, offering significant guidance for the practical deployment of future maritime observation and monitoring missions. Full article

This study investigates the multi-locality and multi-temporal characteristics of mobility destinations in Zanjan, Iran, throughout a typical day. Existing approaches often overlook critical geographical concepts, including the influence of multiple motivational factors on destination choice behavior, the clustering of destinations, and the spatiotemporal dynamics of preferred destinations. To address these gaps, Agent-Based Modeling (ABM) was employed to simulate individual daily flows to preferred destinations. An integrated pattern recognition approach combining machine learning clustering (k-means), hotspot analysis, and 3D mapping was utilized to facilitate visual analytics of individual destination choices, with special emphasis on applications for transportation planning. Four optimal destination clusters were identified, with hotspot analysis revealing a concentration of preferred destinations in Cluster 1, located within the Central Business District (CBD), suggesting a monocentric spatial structure. Temporal analysis demonstrated that destination clusters exhibit dynamic spatial and temporal changes over the course of the day. These findings provide new insights into managing travel behavior and offer practical implications for urban planning and transportation policy regarding individuals’ daily movement strategies. Full article

This work proposes a multiple unmanned aerial vehicles (UAVs) cooperative trajectory planning scheme constructed by multi-agent reinforcement learning with hybrid critics, improving the searching and tracking efficiency and fairness when the dynamic unmanned surface vehicle (USV) swarm exceeds the number of UAVs. A confidence map of targets’ existence probability with spatio-temporal decay is first established through a local information fusion mechanism based on Bayesian update theory. It leads to a reformulation of the problem model into a communication-enhanced partially observable Markov decision process. To suppress policy variance and credibility imbalance of the multi-UAVs, a center-sub-critics deep deterministic policy gradient algorithm is then proposed, combining multiple centralized critics with decentralized critics. Meanwhile, a segmented reward function is designed to incentivize the UAV to revisit detected targets. Finally, the simulation results compared with diverse baseline algorithms demonstrate the efficacy and scalability of the proposed scheme in this paper. Full article

This paper presents PLAM (Plovdiv Air Monitoring)—a regional multi-agent platform for air quality monitoring, semantic reasoning, and forecasting. The platform uses a hybrid architecture that combines two types of intelligent agents: classic BDI (Belief-Desire-Intention) agents for complex, goal-oriented behavior and planning, and ReAct agents based on large language models (LLM) for quick response, analysis, and interaction with users. The system integrates data from heterogeneous sources, including local IoT sensor networks and public external services, enriching it with a specialized OWL ontology of environmental norms. Based on this data, the platform performs comparative analysis, detection of anomalies and inconsistencies between measurements, as well as predictions using machine learning models. The results are visualized and presented to users via a web interface and mobile application, including personalized alerts and recommendations. The architecture demonstrates essential properties of an intelligent agent such as autonomy, proactivity, reactivity, and social capabilities. The implementation and testing in the city of Plovdiv demonstrate the system’s ability to provide a more objective and comprehensive assessment of air quality, revealing significant differences between measurements from different institutions. The platform offers a modular and adaptive design, making it applicable to other regions, and outlines future development directions, such as creating a specialized small language model and expanding sensor capabilities. Full article

A hybrid AI and LLM-enabled architecture is presented for real-time decision support in industrial batch processes, where supervision still relies heavily on human operators and ad hoc SCADA logic. Unlike algorithmic contributions proposing novel AI methods, this work addresses the practical integration and deployment challenges arising when applying existing AI techniques to safety-critical industrial environments with legacy PLC/SCADA infrastructure and real-time constraints. The framework combines deterministic rule-based agents, fuzzy and statistical enrichment, and large language models (LLMs) to support monitoring, diagnostic interpretation, preventive maintenance planning, and operator interaction with minimal manual intervention. High-frequency sensor streams are collected into rolling buffers per active process instance; deterministic agents compute enriched variables, discrete supervisory states, and rule-based alarms, while an LLM-driven analytics agent answers free-form operator queries over the same enriched datasets through a conversational interface. The architecture is instantiated and deployed in the Clean-in-Place (CIP) system of an industrial beverage plant and evaluated following a case study design aimed at demonstrating architectural feasibility and diagnostic behavior under realistic operating regimes rather than statistical generalization. Three representative multi-stage CIP executions—purposively selected from 24 runs monitored during a six-month deployment—span nominal baseline, preventive-warning, and diagnostic-alert conditions. The study quantifies stage-specification compliance, state-to-specification consistency, and temporal stability of supervisory states, and performs spot-check audits of numerical consistency between language-based summaries and enriched logs. Results in the evaluated CIP deployment show high time within specification in sanitizing stages (100% compliance across the evaluated runs), coherent and mostly stable supervisory states in variable alkaline conditions (state-specification consistency ${\Gamma }_s\ge 0.98$), and data-grounded conversational diagnostics in real time (median numerical error below 3% in audited samples), without altering the existing CIP control logic. These findings suggest that the architecture can be transferred to other industrial cleaning and batch operations by reconfiguring process-specific rules and ontologies, though empirical validation in other process types remains future work. The contribution lies in demonstrating how to bridge the gap between AI theory and industrial practice through careful system architecture, data transformation pipelines, and integration patterns that enable reliable AI-enhanced decision support in production environments, offering a practical path toward AI-assisted process supervision with explainable conversational interfaces that support preventive maintenance decision-making and equipment health monitoring. Full article

The deployment of autonomous robots in construction remains constrained by the complexity and variability of real-world environments. Conventional programming and single-agent approaches lack the adaptability required for dynamic multi-robot operating conditions, underscoring the need for cooperative, learning-based systems. This paper presents an ROS-based modular framework that integrates Multi-Agent Reinforcement Learning (MARL) into a generic 2D simulation and execution pipeline for cooperative mobile robots in construction-oriented digital environments to enable adaptive task allocation and coordinated execution without predefined datasets or manual scheduling. The framework adopts a centralized-training, decentralized-execution (CTDE) scheme based on Multi-Agent Proximal Policy Optimization (MAPPO) and decomposes the system into interchangeable modules for environment modelling, task representation, robot interfaces, and learning, allowing different layouts, task sets, and robot models to be instantiated without redesigning the core architecture. Validation through an ROS-based 2D simulation and real-world experiments using TurtleBot3 robots demonstrated effective task scheduling, adaptive navigation, and cooperative behavior under uncertainty. In simulation, the learned MAPPO policy is benchmarked against non-learning baselines for multi-robot task allocation, and in real-robot experiments, the same policy is evaluated to quantify and discuss the performance gap between simulated and physical execution. Rather than presenting a complete construction-site deployment, this first study focuses on proposing and validating a reusable MARL–ROS framework and digital testbed for multi-robot task allocation in construction-oriented digital environments. The results show that the framework supports effective cooperative task scheduling, adaptive navigation, and logic-consistent behavior, while highlighting practical issues that arise in sim-to-real transfer. Overall, the framework provides a reusable digital foundation and benchmark for studying adaptive and cooperative multi-robot systems in construction-related planning and management contexts. Full article

Large Language Model (LLM) agents depend heavily on multiple external tools such as APIs, databases and computational services to perform complex tasks. However, these tool executions create latency and introduce costs, particularly when agents handle similar queries or workflows. Most current caching methods focus on LLM prompt–response pairs or execution plans and overlook redundancies at the tool level. To address this, we designed a multi-level caching architecture that captures redundancy at both the workflow and tool level. The proposed system integrates four key components: (1) hierarchical caching that operates at both the workflow and tool level to capture coarse and fine-grained redundancies; (2) dependency-aware invalidation using graph-based techniques to maintain consistency when write operations affect cached reads across execution contexts; (3) category-specific time-to-live (TTL) policies tailored to different data types, e.g., weather APIs, user location, database queries and filesystem and computational tasks; and (4) session isolation to ensure multi-tenant cache safety through automatic session scoping. We evaluated the system using synthetic data with 2.25 million queries across ten configurations in fifteen runs. In addition, we conducted four targeted evaluations—write intensity robustness from 4 to 30% writes, personalized memory effects under isolated vs. shared cache modes, workflow-level caching comparison and workload sensitivity across five access distributions—on an additional 2.565 million queries, bringing the total experimental scope to 4.815 million executed queries. The architecture achieved 76.5% caching efficiency, reducing query processing time by 13.3× and lowering estimated costs by 73.3% compared to a no-cache baseline. Multi-tenant testing with fifteen concurrent tenants confirmed robust session isolation and 74.1% efficiency under concurrent workloads. Our evaluation used controlled synthetic workloads following Zipfian distributions, which are commonly used in caching research. While absolute hit rates vary by deployment domain, the architectural principles of hierarchical caching, dependency tracking and session isolation remain broadly applicable. Full article

Path planning for multiple four-way shuttles in high-density warehousing is frequently hampered by efficiency-degrading conflicts, particularly head-on deadlocks. To address this challenge, this paper proposes a multi-agent reinforcement learning (MARL) framework based on Proximal Policy Optimization (PPO). The core of our approach is a novel Cooperative Avoidance Reward Mechanism (CARM), which employs a dual-component reward structure. This structure integrates a distance-guided reward to ensure efficient navigation towards targets and a cooperative avoidance reward that uses both immediate and delayed returns to incentivize implicit collaboration. This design effectively resolves conflicts and mitigates the policy instability often caused by traditional collision penalties. Experiments in a 20 × 20 grid simulation environment demonstrated that, compared to a rule-based A* and Conflict-Based Search (CBS) algorithms, the proposed method reduced the average travel distance and total time by 35.8% and 31.5%, respectively, while increasing system throughput by 49.7% and maintaining a task success rate of over 95%. Ablation studies further confirmed the critical role of CARM in achieving stable multi-agent collaboration. This work offers a scalable and efficient data-driven solution for real-time path planning in complex automated warehousing systems. Full article