The quasi-repetitive nature of construction work and the resulting lack of generalizability in programming construction robots presents persistent challenges to the broad adoption of robots in the construction industry. Robots cannot achieve generalist capabilities as skills learnt from one domain cannot readily transfer to another work domain or be directly used to perform a different set of tasks. Human workers have to arduously reprogram their scene-understanding, path-planning, and manipulation components to enable the robots to perform alternate work tasks. The methods presented in this paper resolve a significant proportion of such reprogramming workload by proposing a generalizable learning architecture that directly teaches robots versatile task-performance skills through crowdsourced online natural language instructions. A Large Language Model (LLM), a standardized and modularized hierarchical modeling approach, and Building Information Modeling-Robot sematic data pipeline are developed to address the multi-task skill transfer problem. The proposed skill standardization scheme and LLM-based hierarchical skill learning framework were tested with a long-horizon drywall installation experiment using a full-scale industrial robotic manipulator. The resulting robot task learning scheme achieves multi-task reprogramming with minimal effort and high quality.
Large language model (LLM) based task plans and corresponding human demonstrations for embodied AI may be noisy, with unnecessary actions, redundant navigation, and logical errors that reduce policy quality. We propose an iterative verification framework in which a Judge LLM critiques action sequences and a Planner LLM applies the revisions, yielding progressively cleaner and more spatially coherent trajectories. Unlike rule-based approaches, our method relies on natural language prompting, enabling broad generalization across error types including irrelevant actions, contradictions, and missing steps. On a set of manually annotated actions from the TEACh embodied AI dataset, our framework achieves up to 90% recall and 100% precision across four state-of-the-art LLMs (GPT o4-mini, DeepSeek-R1, Gemini 2.5, LLaMA 4 Scout). The refinement loop converges quickly, with 96.5% of sequences requiring at most three iterations, while improving both temporal efficiency and spatial action organization. Crucially, the method preserves human error-recovery patterns rather than collapsing them, supporting future work on robust corrective behavior. By establishing plan verification as a reliable LLM capability for spatial planning and action refinement, we provide a scalable path to higher-quality training data for imitation learning in embodied AI.
With advances in large language models (LLMs), researchers are creating new systems that can perform AI-driven analytics over large unstructured datasets. Recent work has explored executing such analytics queries using semantic operators -- a declarative set of AI-powered data transformations with natural language specifications. However, even when optimized, these operators can be expensive to execute on millions of records and their iterator execution semantics make them ill-suited for interactive data analytics tasks. In another line of work, Deep Research systems have demonstrated an ability to answer natural language question(s) over large datasets. These systems use one or more LLM agent(s) to plan their execution, process the dataset(s), and iteratively refine their answer. However, these systems do not explicitly optimize their query plans which can lead to poor plan execution. In order for AI-driven analytics to excel, we need a runtime which combines the optimized execution of semantic operators with the flexibility and more dynamic execution of Deep Research systems. As a first step towards this vision, we build a prototype which enables Deep Research agents to write and execute optimized semantic operator programs. We evaluate our prototype and demonstrate that it can outperform a handcrafted semantic operator program and open Deep Research systems on two basic queries. Compared to a standard open Deep Research agent, our prototype achieves up to 1.95x better F1-score. Furthermore, even if we give the agent access to semantic operators as tools, our prototype still achieves cost and runtime savings of up to 76.8% and 72.7% thanks to its optimized execution.
Effective planning requires strong world models, but high-level world models that can understand and reason about actions with semantic and temporal abstraction remain largely underdeveloped. We introduce the Vision Language World Model (VLWM), a foundation model trained for language-based world modeling on natural videos. Given visual observations, the VLWM first infers the overall goal achievements then predicts a trajectory composed of interleaved actions and world state changes. Those targets are extracted by iterative LLM Self-Refine conditioned on compressed future observations represented by Tree of Captions. The VLWM learns both an action policy and a dynamics model, which respectively facilitates reactive system-1 plan decoding and reflective system-2 planning via cost minimization. The cost evaluates the semantic distance between the hypothetical future states given by VLWM roll-outs and the expected goal state, and is measured by a critic model that we trained in a self-supervised manner. The VLWM achieves state-of-the-art Visual Planning for Assistance (VPA) performance on both benchmark evaluations and our proposed PlannerArena human evaluations, where system-2 improves the Elo score by +27% upon system-1. The VLWM models also outperforms strong VLM baselines on RoboVQA and WorldPrediction benchmark.
The emergence of agentic reinforcement learning (Agentic RL) marks a paradigm shift from conventional reinforcement learning applied to large language models (LLM RL), reframing LLMs from passive sequence generators into autonomous, decision-making agents embedded in complex, dynamic worlds. This survey formalizes this conceptual shift by contrasting the degenerate single-step Markov Decision Processes (MDPs) of LLM-RL with the temporally extended, partially observable Markov decision processes (POMDPs) that define Agentic RL. Building on this foundation, we propose a comprehensive twofold taxonomy: one organized around core agentic capabilities, including planning, tool use, memory, reasoning, self-improvement, and perception, and the other around their applications across diverse task domains. Central to our thesis is that reinforcement learning serves as the critical mechanism for transforming these capabilities from static, heuristic modules into adaptive, robust agentic behavior. To support and accelerate future research, we consolidate the landscape of open-source environments, benchmarks, and frameworks into a practical compendium. By synthesizing over five hundred recent works, this survey charts the contours of this rapidly evolving field and highlights the opportunities and challenges that will shape the development of scalable, general-purpose AI agents.
Language-guided long-horizon manipulation of deformable objects presents significant challenges due to high degrees of freedom, complex dynamics, and the need for accurate vision-language grounding. In this work, we focus on multi-step cloth folding, a representative deformable-object manipulation task that requires both structured long-horizon planning and fine-grained visual perception. To this end, we propose a unified framework that integrates a Large Language Model (LLM)-based planner, a Vision-Language Model (VLM)-based perception system, and a task execution module. Specifically, the LLM-based planner decomposes high-level language instructions into low-level action primitives, bridging the semantic-execution gap, aligning perception with action, and enhancing generalization. The VLM-based perception module employs a SigLIP2-driven architecture with a bidirectional cross-attention fusion mechanism and weight-decomposed low-rank adaptation (DoRA) fine-tuning to achieve language-conditioned fine-grained visual grounding. Experiments in both simulation and real-world settings demonstrate the method's effectiveness. In simulation, it outperforms state-of-the-art baselines by 2.23, 1.87, and 33.3 on seen instructions, unseen instructions, and unseen tasks, respectively. On a real robot, it robustly executes multi-step folding sequences from language instructions across diverse cloth materials and configurations, demonstrating strong generalization in practical scenarios. Project page: https://language-guided.netlify.app/
Large Language Models (LLMs) in agentic workflows combine multi-step reasoning, tool use, and collaboration across multiple specialized agents. Existing LLM serving engines optimize indi- vidual calls in isolation, while multi-agent frameworks focus on orchestration without system-level performance planning. As a result, repeated prompts, overlapping contexts, and concurrent ex- ecutions create substantial redundancy and poor GPU utilization, especially in batch analytics scenarios. We introduce Halo, a system that brings batch query processing and optimization into agentic LLM workflows. Halo represents each workflow as a structured query plan DAG and constructs a consoli- dated graph for batched queries that exposes shared computation. Guided by a cost model that jointly considers prefill and decode costs, cache reuse, and GPU placement, Halo performs plan-level op- timization to minimize redundant execution. Its runtime integrates adaptive batching, KV-cache sharing and migration, along with compute-communication overlap to maximize hardware efficiency. Evaluation across six benchmarks shows that Halo achieves up to 18.6x speedup for batch inference and 4.7x throughput im- provement under online serving, scaling to workloads of tens of thousands of queries and complex graphs. These gains are achieved without compromising output quality. By unifying query optimiza- tion with LLM serving, Halo enables efficient agentic workflows in data analytics and decision-making applications.
Recent advances in large language models (LLMs) have enabled agent-based recommendation systems with strong semantic understanding and flexible reasoning capabilities. While LLM-based agents deployed in the cloud offer powerful personalization, they often suffer from privacy concerns, limited access to real-time signals, and scalability bottlenecks. Conversely, on-device agents ensure privacy and responsiveness but lack the computational power for global modeling and large-scale retrieval. To bridge these complementary limitations, we propose CDA4Rec, a novel Cloud-Device collaborative framework for sequential Recommendation, powered by dual agents: a cloud-side LLM and a device-side small language model (SLM). CDA4Rec tackles the core challenge of cloud-device coordination by decomposing the recommendation task into modular sub-tasks including semantic modeling, candidate retrieval, structured user modeling, and final ranking, which are allocated to cloud or device based on computational demands and privacy sensitivity. A strategy planning mechanism leverages the cloud agent's reasoning ability to generate personalized execution plans, enabling context-aware task assignment and partial parallel execution across agents. This design ensures real-time responsiveness, improved efficiency, and fine-grained personalization, even under diverse user states and behavioral sparsity. Extensive experiments across multiple real-world datasets demonstrate that CDA4Rec consistently outperforms competitive baselines in both accuracy and efficiency, validating its effectiveness in heterogeneous and resource-constrained environments.
Large language models (LLMs) often produce correct answers while relying on flawed or irrelevant reasoning traces, undermining their trustworthiness in high-stakes domains. We propose Counterfactual Sensitivity Regularization (CSR), a lightweight training objective that enforces dependence between intermediate reasoning and final outputs. CSR introduces automated, operator-level counterfactual interventions (e.g., swapping "+" with "-") during training and penalizes models that preserve the same answer under logically invalid traces. This requires only one additional forward pass per sample. To measure faithfulness, we introduce Counterfactual Outcome Sensitivity (COS), which quantifies the impact of such perturbations on model predictions. Across structured reasoning tasks - arithmetic (GSM8K), logical deduction (PrOntoQA), and planning (Blocks World) - CSR improves faithfulness by up to 70 percentage points over standard fine-tuning and process supervision, with only minor accuracy loss. The learned sensitivity generalizes to larger models and synergizes with inference-time methods such as self-consistency. A pilot study on HellaSwag further demonstrates that extending CSR with semantic perturbations can enhance faithfulness in commonsense reasoning.
With the rapid advancement of Large Language Models (LLMs) and Artificial Intelligence (AI) agents, agentic workflows are showing transformative potential in education. This study introduces the Agentic Workflow for Education (AWE), a four-component model comprising self-reflection, tool invocation, task planning, and multi-agent collaboration. We distinguish AWE from traditional LLM-based linear interactions and propose a theoretical framework grounded in the von Neumann Multi-Agent System (MAS) architecture. Through a paradigm shift from static prompt-response systems to dynamic, nonlinear workflows, AWE enables scalable, personalized, and collaborative task execution. We further identify four core application domains: integrated learning environments, personalized AI-assisted learning, simulation-based experimentation, and data-driven decision-making. A case study on automated math test generation shows that AWE-generated items are statistically comparable to real exam questions, validating the model's effectiveness. AWE offers a promising path toward reducing teacher workload, enhancing instructional quality, and enabling broader educational innovation.
As the social environment is growing more complex and collaboration is deepening, factors affecting the healthy development of service ecosystem are constantly changing and diverse, making its governance a crucial research issue. Applying the scenario analysis method and conducting scenario rehearsals by constructing an experimental system before managers make decisions, losses caused by wrong decisions can be largely avoided. However, it relies on predefined rules to construct scenarios and faces challenges such as limited information, a large number of influencing factors, and the difficulty of measuring social elements. These challenges limit the quality and efficiency of generating social and uncertain scenarios for the service ecosystem. Therefore, we propose a scenario generator design method, which adaptively coordinates three Large Language Model (LLM) empowered agents that autonomously optimize experimental schemes to construct an experimental system and generate high quality scenarios. Specifically, the Environment Agent (EA) generates social environment including extremes, the Social Agent (SA) generates social collaboration structure, and the Planner Agent (PA) couples task-role relationships and plans task solutions. These agents work in coordination, with the PA adjusting the experimental scheme in real time by perceiving the states of each agent and these generating scenarios. Experiments on the ProgrammableWeb dataset illustrate our method generates more accurate scenarios more efficiently, and innovatively provides an effective way for service ecosystem governance related experimental system construction.
GUI agents powered by LLMs show promise in interacting with diverse digital environments. Among these, video games offer a valuable testbed due to their varied interfaces, with adventure games posing additional challenges through complex, narrative-driven interactions. Existing game benchmarks, however, lack diversity and rarely evaluate agents on completing entire storylines. To address this, we introduce FlashAdventure, a benchmark of 34 Flash-based adventure games designed to test full story arc completion and tackle the observation-behavior gap: the challenge of remembering and acting on earlier gameplay information. We also propose CUA-as-a-Judge, an automated gameplay evaluator, and COAST, an agentic framework leveraging long-term clue memory to better plan and solve sequential tasks. Experiments show current GUI agents struggle with full story arcs, while COAST improves milestone completion by bridging the observation-behavior gap. Nonetheless, a marked discrepancy between humans and best-performing agents warrants continued research efforts to narrow this divide.
We introduce CoreThink, a state-of-the-art Reasoning Layer built upon a novel reasoning method called General Symbolics. This approach diverges from reasoning paradigms such as test-time scaling, Supervised Fine-Tuning (SFT), and Reinforcement Learning with Verifiable Rewards (RLVR). CoreThink General Symbolic Reasoner (GSR) is specifically structured around three key use cases: tool-calling, code generation, and planning, demonstrating exemplary performance across a total of seven benchmarks in their respective areas. Notably, we are achieving SOTA scores of 66.66% on Livecodebench v6, 89% on Instruction-Following Evals, and 24.4% on ARC-AGI-2. We also present an agentic coding IDE, developed using the principles of General Symbolics, which achieves a state-of-the-art accuracy of 62.3% on SWE-Bench Lite. We are able to achieve these improvements without any fine-tuning or training costs. Our Reasoning Layer is designed to provide a pure performance uplift, ensuring that a model's accuracy on reasoning tasks is never negatively impacted. We argue that incumbent methods will eventually lead to diminishing returns in LLM performance, necessitating the development of new reasoning techniques. This technical report details our approach at a high level and the availability of the CoreThink models for reasoning-intensive use cases.
This paper presents the development of an AI-powered workflow that uses Large Language Models (LLMs) to assist in drafting schematic architectural floor plans from natural language prompts. The proposed system interprets textual input to automatically generate layout options including walls, doors, windows, and furniture arrangements. It combines prompt engineering, a furniture placement refinement algorithm, and Python scripting to produce spatially coherent draft plans compatible with design tools such as Autodesk Revit. A case study of a mid-sized residential layout demonstrates the approach's ability to generate functional and structured outputs with minimal manual effort. The workflow is designed for transparent replication, with all key prompt specifications documented to enable independent implementation by other researchers. In addition, the generated models preserve the full range of Revit-native parametric attributes required for direct integration into professional BIM processes.
The Jump-Jump game, as a simple yet challenging casual game, provides an ideal testing environment for studying LLM decision-making capabilities. The game requires players to precisely control jumping force based on current position and target platform distance, involving multiple cognitive aspects including spatial reasoning, physical modeling, and strategic planning. It illustrates the basic gameplay mechanics of the Jump-Jump game, where the player character (red circle) must jump across platforms with appropriate force to maximize score.
Recent advances in autonomous digital agents from industry (e.g., Manus AI and Gemini's research mode) highlight potential for structured tasks by autonomous decision-making and task decomposition; however, it remains unclear to what extent the agent-based systems can improve multimodal web data productivity. We study this in the realm of journalism, which requires iterative planning, interpretation, and contextual reasoning from multimodal raw contents to form a well structured news. We introduce NEWSAGENT, a benchmark for evaluating how agents can automatically search available raw contents, select desired information, and edit and rephrase to form a news article by accessing core journalistic functions. Given a writing instruction and firsthand data as how a journalist initiates a news draft, agents are tasked to identify narrative perspectives, issue keyword-based queries, retrieve historical background, and generate complete articles. Unlike typical summarization or retrieval tasks, essential context is not directly available and must be actively discovered, reflecting the information gaps faced in real-world news writing. NEWSAGENT includes 6k human-verified examples derived from real news, with multimodal contents converted to text for broad model compatibility. We evaluate open- and closed-sourced LLMs with commonly-used agentic frameworks on NEWSAGENT, which shows that agents are capable of retrieving relevant facts but struggling with planning and narrative integration. We believe that NEWSAGENT serves a realistic testbed for iterating and evaluating agent capabilities in terms of multimodal web data manipulation to real-world productivity.
The emergence of large-language models (LLMs) has enabled a new class of semantic data processing systems (SDPSs) to support declarative queries against unstructured documents. Existing SDPSs are, however, lacking a unified algebraic foundation, making their queries difficult to compose, reason, and optimize. We propose a new semantic algebra, SABER (Semantic Algebra Based on Extended Relational algebra), opening the possibility of semantic operations' logical plan construction, optimization, and formal correctness guarantees. We further propose to implement SABER in a SQL-compatible syntax so that it natively supports mixed structured/unstructured data processing. With SABER, we showcase the feasibility of providing a unified interface for existing SDPSs so that it can effectively mix and match any semantically-compatible operator implementation from any SDPS, greatly enhancing SABER's applicability for community contributions.
Introduction: Large language models (LLM) have shown great potential in clinical decision support. GPT-5 is a novel LLM system that has been specifically marketed towards oncology use. Methods: Performance was assessed using two complementary benchmarks: (i) the ACR Radiation Oncology In-Training Examination (TXIT, 2021), comprising 300 multiple-choice items, and (ii) a curated set of 60 authentic radiation oncologic vignettes representing diverse disease sites and treatment indications. For the vignette evaluation, GPT-5 was instructed to generate concise therapeutic plans. Four board-certified radiation oncologists rated correctness, comprehensiveness, and hallucinations. Inter-rater reliability was quantified using Fleiss' \k{appa}. Results: On the TXIT benchmark, GPT-5 achieved a mean accuracy of 92.8%, outperforming GPT-4 (78.8%) and GPT-3.5 (62.1%). Domain-specific gains were most pronounced in Dose and Diagnosis. In the vignette evaluation, GPT-5's treatment recommendations were rated highly for correctness (mean 3.24/4, 95% CI: 3.11-3.38) and comprehensiveness (3.59/4, 95% CI: 3.49-3.69). Hallucinations were rare with no case reaching majority consensus for their presence. Inter-rater agreement was low (Fleiss' \k{appa} 0.083 for correctness), reflecting inherent variability in clinical judgment. Errors clustered in complex scenarios requiring precise trial knowledge or detailed clinical adaptation. Discussion: GPT-5 clearly outperformed prior model variants on the radiation oncology multiple-choice benchmark. Although GPT-5 exhibited favorable performance in generating real-world radiation oncology treatment recommendations, correctness ratings indicate room for further improvement. While hallucinations were infrequent, the presence of substantive errors underscores that GPT-5-generated recommendations require rigorous expert oversight before clinical implementation.
This paper presents an integrated framework that combines traditional network optimization models with large language models (LLMs) to deliver interactive, explainable, and role-aware decision support for supply chain planning. The proposed system bridges the gap between complex operations research outputs and business stakeholder understanding by generating natural language summaries, contextual visualizations, and tailored key performance indicators (KPIs). The core optimization model addresses tactical inventory redistribution across a network of distribution centers for multi-period and multi-item, using a mixed-integer formulation. The technical architecture incorporates AI agents, RESTful APIs, and a dynamic user interface to support real-time interaction, configuration updates, and simulation-based insights. A case study demonstrates how the system improves planning outcomes by preventing stockouts, reducing costs, and maintaining service levels. Future extensions include integrating private LLMs, transfer learning, reinforcement learning, and Bayesian neural networks to enhance explainability, adaptability, and real-time decision-making.
Proactive dialogue has emerged as a critical and challenging research problem in advancing large language models (LLMs). Existing works predominantly focus on domain-specific or task-oriented scenarios, which leads to fragmented evaluations and limits the comprehensive exploration of models' proactive conversation abilities. In this work, we propose ProactiveEval, a unified framework designed for evaluating proactive dialogue capabilities of LLMs. This framework decomposes proactive dialogue into target planning and dialogue guidance, establishing evaluation metrics across various domains. Moreover, it also enables the automatic generation of diverse and challenging evaluation data. Based on the proposed framework, we develop 328 evaluation environments spanning 6 distinct domains. Through experiments with 22 different types of LLMs, we show that DeepSeek-R1 and Claude-3.7-Sonnet exhibit exceptional performance on target planning and dialogue guidance tasks, respectively. Finally, we investigate how reasoning capabilities influence proactive behaviors and discuss their implications for future model development.
Recent advances in reasoning and planning capabilities of large language models (LLMs) have enabled their potential as autonomous agents capable of tool use in dynamic environments. However, in multi-turn conversational environments like $\tau$-bench, these agents often struggle with consistent reasoning, adherence to domain-specific policies, and extracting correct information over a long horizon of tool-calls and conversation. To capture and mitigate these failures, we conduct a comprehensive manual analysis of the common errors occurring in the conversation trajectories. We then experiment with reformulations of inputs to the tool-calling agent for improvement in agent decision making. Finally, we propose the Input-Reformulation Multi-Agent (IRMA) framework, which automatically reformulates user queries augmented with relevant domain rules and tool suggestions for the tool-calling agent to focus on. The results show that IRMA significantly outperforms ReAct, Function Calling, and Self-Reflection by 16.1%, 12.7%, and 19.1%, respectively, in overall pass^5 scores. These findings highlight the superior reliability and consistency of IRMA compared to other methods in dynamic environments.
Large language models (LLMs) have become essential for applications such as text summarization, sentiment analysis, and automated question-answering. Recently, LLMs have also been integrated into relational database management systems to enhance querying and support advanced data processing. Companies such as Amazon, Databricks, Google, and Snowflake offer LLM invocation directly within SQL, denoted as LLM queries, to boost data insights. However, open-source solutions currently have limited functionality and poor performance. In this work, we present an early exploration of two open-source systems and one enterprise platform, using five representative queries to expose functional, performance, and scalability limits in today's SQL-invoked LLM integrations. We identify three main issues: enforcing structured outputs, optimizing resource utilization, and improving query planning. We implemented initial solutions and observed improvements in accommodating LLM powered SQL queries. These early gains demonstrate that tighter integration of LLM+DBMS is the key to scalable and efficient processing of LLM queries.
We introduce MCP-Bench, a benchmark for evaluating large language models (LLMs) on realistic, multi-step tasks that demand tool use, cross-tool coordination, precise parameter control, and planning/reasoning for solving tasks. Built on the Model Context Protocol (MCP), MCP-Bench connects LLMs to 28 representative live MCP servers spanning 250 tools across domains such as finance, traveling, scientific computing, and academic search. Unlike prior API-based benchmarks, each MCP server provides a set of complementary tools designed to work together, enabling the construction of authentic, multi-step tasks with rich input-output coupling. Tasks in MCP-Bench test agents' ability to retrieve relevant tools from fuzzy instructions without explicit tool names, plan multi-hop execution trajectories for complex objectives, ground responses in intermediate tool outputs, and orchestrate cross-domain workflows - capabilities not adequately evaluated by existing benchmarks that rely on explicit tool specifications, shallow few-step workflows, and isolated domain operations. We propose a multi-faceted evaluation framework covering tool-level schema understanding and usage, trajectory-level planning, and task completion. Experiments on 20 advanced LLMs reveal persistent challenges in MCP-Bench. Code and data: https://github.com/Accenture/mcp-bench.
Using automated reasoning, code synthesis, and contextual decision-making, we introduce a new threat that exploits large language models (LLMs) to autonomously plan, adapt, and execute the ransomware attack lifecycle. Ransomware 3.0 represents the first threat model and research prototype of LLM-orchestrated ransomware. Unlike conventional malware, the prototype only requires natural language prompts embedded in the binary; malicious code is synthesized dynamically by the LLM at runtime, yielding polymorphic variants that adapt to the execution environment. The system performs reconnaissance, payload generation, and personalized extortion, in a closed-loop attack campaign without human involvement. We evaluate this threat across personal, enterprise, and embedded environments using a phase-centric methodology that measures quantitative fidelity and qualitative coherence in each attack phase. We show that open source LLMs can generate functional ransomware components and sustain closed-loop execution across diverse environments. Finally, we present behavioral signals and multi-level telemetry of Ransomware 3.0 through a case study to motivate future development of better defenses and policy enforcements to address novel AI-enabled ransomware attacks.
Recent studies have explored integrating Large Language Models (LLMs) with search engines to leverage both the LLMs' internal pre-trained knowledge and external information. Specially, reinforcement learning (RL) has emerged as a promising paradigm for enhancing LLM reasoning through multi-turn interactions with search engines. However, existing RL-based search agents rely on a single LLM to handle both search planning and question-answering (QA) tasks in an end-to-end manner, which limits their ability to optimize both capabilities simultaneously. In practice, sophisticated AI search systems often employ a large, frozen LLM (e.g., GPT-4, DeepSeek-R1) to ensure high-quality QA. Thus, a more effective and efficient approach is to utilize a small, trainable LLM dedicated to search planning. In this paper, we propose \textbf{AI-SearchPlanner}, a novel reinforcement learning framework designed to enhance the performance of frozen QA models by focusing on search planning. Specifically, our approach introduces three key innovations: 1) Decoupling the Architecture of the Search Planner and Generator, 2) Dual-Reward Alignment for Search Planning, and 3) Pareto Optimization of Planning Utility and Cost, to achieve the objectives. Extensive experiments on real-world datasets demonstrate that AI SearchPlanner outperforms existing RL-based search agents in both effectiveness and efficiency, while exhibiting strong generalization capabilities across diverse frozen QA models and data domains.
Cataract surgery remains one of the most widely performed and effective procedures for vision restoration. Effective surgical planning requires integrating diverse clinical examinations for patient assessment, intraocular lens (IOL) selection, and risk evaluation. Large language models (LLMs) have shown promise in supporting clinical decision-making. However, existing LLMs often lack the domain-specific expertise to interpret heterogeneous ophthalmic data and provide actionable surgical plans. To enhance the model's ability to interpret heterogeneous ophthalmic reports, we propose a knowledge-driven Multi-Agent System (MAS), where each agent simulates the reasoning process of specialist ophthalmologists, converting raw clinical inputs into structured, actionable summaries in both training and deployment stages. Building on MAS, we introduce CataractSurg-80K, the first large-scale benchmark for cataract surgery planning that incorporates structured clinical reasoning. Each case is annotated with diagnostic questions, expert reasoning chains, and structured surgical recommendations. We further introduce Qwen-CSP, a domain-specialized model built on Qwen-4B, fine-tuned through a multi-stage process tailored for surgical planning. Comprehensive experiments show that Qwen-CSP outperforms strong general-purpose LLMs across multiple metrics. Our work delivers a high-quality dataset, a rigorous benchmark, and a domain-adapted LLM to facilitate future research in medical AI reasoning and decision support.
The functionality of Large Language Model (LLM) agents is primarily determined by two capabilities: action planning and answer summarization. The former, action planning, is the core capability that dictates an agent's performance. However, prevailing training paradigms employ end-to-end, multi-objective optimization that jointly trains both capabilities. This paradigm faces two critical challenges: imbalanced optimization objective allocation and scarcity of verifiable data, making it difficult to enhance the agent's planning capability. To address these challenges, we propose Reinforcement Learning with Tool-use Rewards (RLTR), a novel framework that decouples the training process to enable a focused, single-objective optimization of the planning module. Crucially, RLTR introduces a reward signal based on tool-use completeness to directly evaluate the quality of tool invocation sequences. This method offers a more direct and reliable training signal than assessing the final response content, thereby obviating the need for verifiable data. Our experiments demonstrate that RLTR achieves an 8%-12% improvement in planning performance compared to end-to-end baselines. Moreover, this enhanced planning capability, in turn, translates to a 5%-6% increase in the final response quality of the overall agent system.
Noisy Intermediate-Scale Quantum (NISQ) devices have begun to exhibit early quantum advantages on classically intractable problems, spanning physics simulations to Gaussian boson sampling. Yet, realizing these benefits remains challenging for non-experts, primarily due to the complexities of programming in Open Quantum Assembly Language (OpenQASM). Although Large Language Model (LLM)-based agents have shown promise in automating classical programming workflows, their quantum counterparts have largely been restricted to specialized tasks such as quantum chemistry or error correction. In this paper, we present QAgent, an LLM-powered multi-agent system that fully automates OpenQASM programming. By integrating task planning, in-context few-shot learning, retrieval-augmented generation (RAG) for long-term context, predefined generation tools, and chain-of-thought (CoT) reasoning, the agents systematically improve both compilation and functional correctness. Our evaluations demonstrate substantial improvements: across multiple LLMs of varying sizes, QAgent enhances the accuracy of QASM code generation by 71.6\% compared to previous static LLM-based approaches. We envision this multi-agent system as a key enabler for democratizing quantum programming, bridging expertise gaps, and accelerating the practical adoption of quantum computing.
We present DELIVER (Directed Execution of Language-instructed Item Via Engineered Relay), a fully integrated framework for cooperative multi-robot pickup and delivery driven by natural language commands. DELIVER unifies natural language understanding, spatial decomposition, relay planning, and motion execution to enable scalable, collision-free coordination in real-world settings. Given a spoken or written instruction, a lightweight instance of LLaMA3 interprets the command to extract pickup and delivery locations. The environment is partitioned using a Voronoi tessellation to define robot-specific operating regions. Robots then compute optimal relay points along shared boundaries and coordinate handoffs. A finite-state machine governs each robot's behavior, enabling robust execution. We implement DELIVER on the MultiTRAIL simulation platform and validate it in both ROS2-based Gazebo simulations and real-world hardware using TurtleBot3 robots. Empirical results show that DELIVER maintains consistent mission cost across varying team sizes while reducing per-agent workload by up to 55% compared to a single-agent system. Moreover, the number of active relay agents remains low even as team size increases, demonstrating the system's scalability and efficient agent utilization. These findings underscore DELIVER's modular and extensible architecture for language-guided multi-robot coordination, advancing the frontiers of cyber-physical system integration.
Large language model (LLM)-based agents have demonstrated remarkable capabilities in decision-making tasks, but struggle significantly with complex, long-horizon planning scenarios. This arises from their lack of macroscopic guidance, causing disorientation and failures in complex tasks, as well as insufficient continuous oversight during execution, rendering them unresponsive to environmental changes and prone to deviations. To tackle these challenges, we introduce HiPlan, a hierarchical planning framework that provides adaptive global-local guidance to boost LLM-based agents'decision-making. HiPlan decomposes complex tasks into milestone action guides for general direction and step-wise hints for detailed actions. During the offline phase, we construct a milestone library from expert demonstrations, enabling structured experience reuse by retrieving semantically similar tasks and milestones. In the execution phase, trajectory segments from past milestones are dynamically adapted to generate step-wise hints that align current observations with the milestone objectives, bridging gaps and correcting deviations. Extensive experiments across two challenging benchmarks demonstrate that HiPlan substantially outperforms strong baselines, and ablation studies validate the complementary benefits of its hierarchical components.