Large language models (LLMs) can now translate a researcher's plain-language goal into executable computation, yet scientific workflows demand determinism, provenance, and governance that are difficult to guarantee when an LLM decides what runs. Semi-structured interviews with 18 experts across 10 industrial R&D stakeholders surface 2 competing requirements--deterministic, constrained execution and conversational flexibility without workflow rigidity--together with boundary properties (human-in-the-loop control and transparency) that any resolution must satisfy. We propose schema-gated orchestration as the resolving principle: the schema becomes a mandatory execution boundary at the composed-workflow level, so that nothing runs unless the complete action--including cross-step dependencies--validates against a machine-checkable specification. We operationalize the 2 requirements as execution determinism (ED) and conversational flexibility (CF), and use these axes to review 20 systems spanning 5 architectural groups along a validation-scope spectrum. Scores are assigned via a multi-model protocol--15 independent sessions across 3 LLM families--yielding substantial-to-near-perfect inter-model agreement (Krippendorff a=0.80 for ED and a=0.98 for CF), demonstrating that multi-model LLM scoring can serve as a reusable alternative to human expert panels for architectural assessment. The resulting landscape reveals an empirical Pareto front--no reviewed system achieves both high flexibility and high determinism--but a convergence zone emerges between the generative and workflow-centric extremes. We argue that a schema-gated architecture, separating conversational from execution authority, is positioned to decouple this trade-off, and distill 3 operational principles--clarification-before-execution, constrained plan-act orchestration, and tool-to-workflow-level gating--to guide adoption.
This study investigates the use of large language models (LLMs) for story point estimation. Story points are unitless, project-specific effort estimates that help developers on the scrum team forecast which product backlog items they plan to complete in a sprint. To facilitate this process, machine learning models, especially deep neural networks, have been applied to predict the story points based on the title and description of each item. However, such machine learning models require sufficient amounts of training data (with ground truth story points annotated by human developers) from the same software project to achieve decent prediction performance. This motivated us to explore whether LLMs are capable of (RQ1) predicting story points without training data or (RQ2) with only a few training data points. Our empirical results with four LLMs on 16 software projects show that, without any training data (zero-shot prompting), LLMs can predict story points better than supervised deep learning models trained on 80% of the data. The prediction performance of LLMs can be further improved with a few training examples (few-shot prompting). In addition, a recent study explored the use of comparative judgments (between a given pair of items which one requires more effort to implement) instead of directly annotating the story points to reduce the cognitive burden on developers. Therefore, this study also explores (RQ3) whether comparative judgments are easier to predict than story points for LLMs and (RQ4) whether comparative judgments can serve as few-shot examples for LLMs to improve their predictions of story points. Empirical results suggest that it is not easier for LLMs to predict comparative judgments than to directly estimate the story points, but comparative judgments can serve as few-shot examples to improve the LLMs' prediction performance as well as the human-annotated story points.
Reinforcement learning is a framework for learning to act sequentially in an unknown environment. We propose a natural approach for modeling policy structure in policy gradients. The key idea is to optimize for a subset of future rewards: smaller subsets represent simpler policies, which can be learned more reliably because their empirical gradient estimates are more accurate. Our approach allows for modeling and comparison of different policy classes, including full planning, greedy, K-step lookahead, and segment policies. We evaluate the policies empirically on multiple persona-alignment conversational problems. Different policies excel in different problems, reflecting their different characteristics and highlighting the importance of our studied policy class.
Task planning, the problem of sequencing actions to reach a goal from an initial state, is a core capability requirement for autonomous robotic systems. Whether large language models (LLMs) can serve as viable planners alongside classical symbolic methods remains an open question. We present PyPDDLEngine, an open-source Planning Domain Definition Language (PDDL) simulation engine that exposes planning operations as LLM tool calls through a Model Context Protocol (MCP) interface. Rather than committing to a complete action sequence upfront, the LLM acts as an interactive search policy that selects one action at a time, observes each resulting state, and can reset and retry. We evaluate four approaches on 102 International Planning Competition (IPC) Blocksworld instances under a uniform 180-second budget: Fast Downward lama-first and seq-sat-lama-2011 as classical baselines, direct LLM planning (Claude Haiku 4.5), and agentic LLM planning via PyPDDLEngine. Fast Downward achieves 85.3% success. The direct and agentic LLM approaches achieve 63.7% and 66.7%, respectively, a consistent but modest three-percentage-point advantage for the agentic approach at $5.7\times$ higher token cost per solution. Across most co-solved difficulty blocks, both LLM approaches produce shorter plans than seq-sat-lama-2011 despite its iterative quality improvement, a result consistent with training-data recall rather than generalisable planning. These results suggest that agentic gains depend on the nature of environmental feedback. Coding agents benefit from externally grounded signals such as compiler errors and test failures, whereas PDDL step feedback is self-assessed, leaving the agent to evaluate its own progress without external verification.
Large Language Models (LLMs) excel at multi-step reasoning, yet increasing the structural complexity of inference does not consistently improve system-level returns. Methods such as Tree of Thoughts (ToT), Graph of Thoughts (GoT), and Adaptive Graph of Thoughts (AGoT) can boost accuracy on some benchmarks, but often introduce substantial overhead in token consumption and latency, and their gains can be unstable across task distributions-sometimes underperforming simpler Chain-of-Thought (CoT) or direct input-output prompting (IO). We attribute this inefficiency to stage-wise and node-wise heterogeneity inside GoT-style reasoning pipelines: high-quality planning and final synthesis are globally coupled and typically benefit from strong models, whereas many intermediate subtasks are localized and can be solved accurately by lighter models with far fewer tokens. Motivated by these observations, we propose RouteGoT, a budget-controllable, node-adaptive routing framework for graph-structured reasoning. RouteGoT performs in-graph routing by prioritizing strong models for planning and synthesis, while dynamically allocating lightweight models and cost-effective strategies to leaf subtasks based on predicted difficulty. It further integrates explicit budget constraints into a global inference scheduler to control graph expansion under a user-specified token budget, enabling predictable performance-cost trade-offs. Experiments across reasoning, retrieval, and multi-hop QA benchmarks show that RouteGoT matching or improving accuracy while substantially reducing token usage; specifically, it achieves an average 8.1 percentage points accuracy improvement and 79.1\% output token reduction compared to AGoT. Furthermore, RouteGoT outperforms existing routing baselines by maintaining a superior cost-accuracy trade-off, demonstrating improved robustness under varying budget targets and tasks.
Recent advances in large language models (LLMs) have enabled agentic systems for sequential decision-making. Such agents must perceive their environment, reason across multiple time steps, and take actions that optimize long-term objectives. However, existing web agents struggle on complex, long-horizon tasks due to limited in-context memory for tracking history, weak planning abilities, and greedy behaviors that lead to premature termination. To address these challenges, we propose STRUCTUREDAGENT, a hierarchical planning framework with two core components: (1) an online hierarchical planner that uses dynamic AND/OR trees for efficient search and (2) a structured memory module that tracks and maintains candidate solutions to improve constraint satisfaction in information-seeking tasks. The framework also produces interpretable hierarchical plans, enabling easier debugging and facilitating human intervention when needed. Our results on WebVoyager, WebArena, and custom shopping benchmarks show that STRUCTUREDAGENT improves performance on long-horizon web-browsing tasks compared to standard LLM-based agents.
This demonstration presents U-Parking, a distributed Ultra-Wideband (UWB)-assisted autonomous parking system. By integrating Large Language Models (LLMs)-assisted planning with robust fusion localization and trajectory tracking, it enables reliable automated parking in challenging indoor environments, as validated through real-vehicle demonstrations.
Sequential LLM agents fail on long-horizon planning with hard constraints like budgets and diversity requirements. As planning progresses and context grows, these agents drift from global constraints. We propose HiMAP-Travel, a hierarchical multi-agent framework that splits planning into strategic coordination and parallel day-level execution. A Coordinator allocates resources across days, while Day Executors plan independently in parallel. Three key mechanisms enable this: a transactional monitor enforcing budget and uniqueness constraints across parallel agents, a bargaining protocol allowing agents to reject infeasible sub-goals and trigger re-planning, and a single policy trained with GRPO that powers all agents through role conditioning. On TravelPlanner, HiMAP-Travel with Qwen3-8B achieves 52.78% validation and 52.65% test Final Pass Rate (FPR). In a controlled comparison with identical model, training, and tools, it outperforms the sequential DeepTravel baseline by +8.67~pp. It also surpasses ATLAS by +17.65~pp and MTP by +10.0~pp. On FlexTravelBench multi-turn scenarios, it achieves 44.34% (2-turn) and 37.42% (3-turn) FPR while reducing latency 2.5x through parallelization.
Python bindings are a critical bridge between high-performance C++ libraries and the flexibility of Python, enabling rapid prototyping, reproducible experiments, and integration with simulation and learning frameworks in robotics research. Yet, generating bindings for large codebases is a tedious process that creates a heavy burden for a small group of maintainers. In this work, we investigate the use of Large Language Models (LLMs) to assist in generating nanobind wrappers, with human experts kept in the loop. Our workflow mirrors the structure of the C++ codebase, scaffolds empty wrapper files, and employs LLMs to fill in binding definitions. Experts then review and refine the generated code to ensure correctness, compatibility, and performance. Through a case study on a large C++ motion planning library, we document common failure modes, including mismanaging shared pointers, overloads, and trampolines, and show how in-context examples and careful prompt design improve reliability. Experiments demonstrate that the resulting bindings achieve runtime performance comparable to legacy solutions. Beyond this case study, our results provide general lessons for applying LLMs to binding generation in large-scale C++ projects.
Accurate channel estimation is essential for massive multiple-input multiple-output (MIMO) technologies in next-generation wireless communications. Recently, the radio radiance field (RRF) has emerged as a promising approach for wireless channel modeling, offering a comprehensive spatial representation of channels based on environmental geometry. State-of-the-art RRF reconstruction methods, such as RF-3DGS, can render channel parameters, including gain, angle of arrival, angle of departure, and delay, within milliseconds. However, creating the required 3D environment typically demands precise measurements and advanced computer vision techniques, limiting accessibility. This paper introduces a locally deployable tool that simplifies 3D environment creation for RRF reconstruction. The system combines finetuned language models, generative 3D modeling frameworks, and Blender integration to enable intuitive, chat-based scene design. Specifically, T5-mini is finetuned for parsing user commands, while all-MiniLM-L6-v2 supports semantic retrieval from a local object library. For model generation, LLaMA-Mesh provides fast mesh creation, and Shap-E delivers high-quality outputs. A custom Blender export plugin ensures compatibility with the RF-3DGS pipeline. We demonstrate the tool by constructing 3D models of the NIST lobby and the UW-Madison wireless lab, followed by corresponding RRF reconstructions. This approach significantly reduces modeling complexity, enhancing the usability of RRF for wireless research and spectrum planning.
Procedural planning aims to predict a sequence of actions that transforms an initial visual state into a desired goal, a fundamental ability for intelligent agents operating in complex environments. Existing approaches typically rely on large-scale models that learn procedural structures implicitly, resulting in limited sample-efficiency and high computational cost. In this work we introduce ViterbiPlanNet, a principled framework that explicitly integrates procedural knowledge into the learning process through a Differentiable Viterbi Layer (DVL). The DVL embeds a Procedural Knowledge Graph (PKG) directly with the Viterbi decoding algorithm, replacing non-differentiable operations with smooth relaxations that enable end-to-end optimization. This design allows the model to learn through graph-based decoding. Experiments on CrossTask, COIN, and NIV demonstrate that ViterbiPlanNet achieves state-of-the-art performance with an order of magnitude fewer parameters than diffusion- and LLM-based planners. Extensive ablations show that performance gains arise from our differentiable structure-aware training rather than post-hoc refinement, resulting in improved sample efficiency and robustness to shorter unseen horizons. We also address testing inconsistencies establishing a unified testing protocol with consistent splits and evaluation metrics. With this new protocol, we run experiments multiple times and report results using bootstrapping to assess statistical significance.
World models are essential for planning and evaluation in agentic systems, yet existing approaches lie at two extremes: hand-engineered simulators that offer consistency and reproducibility but are costly to adapt, and implicit neural models that are flexible but difficult to constrain, verify, and debug over long horizons. We seek a principled middle ground that combines the reliability of explicit simulators with the flexibility of learned models, allowing world models to be adapted during online execution. By targeting a broad class of environments whose dynamics are governed by the ordering, timing, and causality of discrete events, such as queueing and service operations, embodied task planning, and message-mediated multi-agent coordination, we advocate explicit, executable discrete-event world models synthesized directly from natural-language specifications. Our approach adopts the DEVS formalism and introduces a staged LLM-based generation pipeline that separates structural inference of component interactions from component-level event and timing logic. To evaluate generated models without a unique ground truth, simulators emit structured event traces that are validated against specification-derived temporal and semantic constraints, enabling reproducible verification and localized diagnostics. Together, these contributions produce world models that are consistent over long-horizon rollouts, verifiable from observable behavior, and efficient to synthesize on demand during online execution.
Robot planning in partially observable environments, where not all objects are known or visible, is a challenging problem, as it requires reasoning under uncertainty through partially observable Markov decision processes. During the execution of a computed plan, a robot may unexpectedly observe task-irrelevant objects, which are typically ignored by naive planners. In this work, we propose incorporating two types of common-sense knowledge: (1) certain objects are more likely to be found in specific locations; and (2) similar objects are likely to be co-located, while dissimilar objects are less likely to be found together. Manually engineering such knowledge is complex, so we explore leveraging the powerful common-sense reasoning capabilities of large language models (LLMs). Our planning and execution framework, CoCo-TAMP, introduces a hierarchical state estimation that uses LLM-guided information to shape the belief over task-relevant objects, enabling efficient solutions to long-horizon task and motion planning problems. In experiments, CoCo-TAMP achieves an average reduction of 62.7% in planning and execution time in simulation, and 72.6% in real-world demonstrations, compared to a baseline that does not incorporate either type of common-sense knowledge.
Automated design of chemical formulations is a cornerstone of materials science, yet it requires navigating a high-dimensional combinatorial space involving discrete compositional choices and continuous geometric constraints. Existing Large Language Model (LLM) agents face significant challenges in this setting, including context window limitations during long-horizon reasoning and path-dependent exploration that may lead to mode collapse. To address these issues, we introduce AI4S-SDS, a closed-loop neuro-symbolic framework that integrates multi-agent collaboration with a tailored Monte Carlo Tree Search (MCTS) engine. We propose a Sparse State Storage mechanism with Dynamic Path Reconstruction, which decouples reasoning history from context length and enables arbitrarily deep exploration under fixed token budgets. To reduce local convergence and improve coverage, we implement a Global--Local Search Strategy: a memory-driven planning module adaptively reconfigures the search root based on historical feedback, while a Sibling-Aware Expansion mechanism promotes orthogonal exploration at the node level. Furthermore, we bridge symbolic reasoning and physical feasibility through a Differentiable Physics Engine, employing a hybrid normalized loss with sparsity-inducing regularization to optimize continuous mixing ratios under thermodynamic constraints. Empirical results show that AI4S-SDS achieves full validity under the adopted HSP-based physical constraints and substantially improves exploration diversity compared to baseline agents. In preliminary lithography experiments, the framework identifies a novel photoresist developer formulation that demonstrates competitive or superior performance relative to a commercial benchmark, highlighting the potential of diversity-driven neuro-symbolic search for scientific discovery.
Recent advances in large language models (LLMs) transform how machine learning (ML) pipelines are developed and evaluated. LLMs enable a new type of workload, agentic pipeline search, in which autonomous or semi-autonomous agents generate, validate, and optimize complete ML pipelines. These agents predominantly operate over popular Python ML libraries and exhibit highly exploratory behavior. This results in thousands of executions for data profiling, pipeline generation, and iterative refinement of pipeline stages. However, the existing Python-based ML ecosystem is built around libraries such as Pandas and scikit-learn, which are designed for human-centric, interactive, sequential workflows and remain constrained by Python's interpretive execution model, library-level isolation, and limited runtime support for executing large numbers of pipelines. Meanwhile, many high-performance ML systems proposed by the systems community either target narrow workload classes or require specialized programming models, which limits their integration with the Python ML ecosystem and makes them largely ill-suited for LLM-based agents. This growing mismatch exposes a fundamental systems challenge in supporting agentic pipeline search at scale. We therefore propose stratum, a unified system infrastructure that decouples pipeline execution from planning and reasoning during agentic pipeline search. Stratum integrates seamlessly with existing Python libraries, compiles batches of pipelines into optimized execution graphs, and efficiently executes them across heterogeneous backends, including a novel Rust-based runtime. We present stratum's architectural vision along with an early prototype, discuss key design decisions, and outline open challenges and research directions. Finally, preliminary experiments show that stratum can significantly speed up large-scale agentic pipeline search up to 16.6x.
Discrete biological sequence optimization requires iterative refinement under strict syntactic constraints. Diffusion models offer progressive refinement but do not naturally expose controllable discrete edit operations, while autoregressive LLMs often lack explicit long-horizon planning for constrained edits. We propose STRIDE (Sequence Trajectory Refinement via Internalized Denoising Emulation), a post-training framework that trains an LLM to emit executable trajectories of atomic edits (INSERT/DELETE/REPLACE) as a verifiable reasoning trace for variable-length refinement. STRIDE combines supervised fine-tuning on Levenshtein-aligned shortest edit demonstrations with group-based policy optimization to align edit trajectories with task rewards while preserving coherent editing behavior. Across protein fluorescence and instruction-conditioned molecular optimization, STRIDE improves variable-length protein editing success from 42% to 89% while increasing novelty from 47% to 97%, and yields stronger validity and controllability compared to diverse baselines. The code is published at https://github.com/daiheng-zhang/STRIDE.
Can a multimodal language model learn to manipulate physical objects by reasoning about its own failures-without gradient updates, demonstrations, or reward engineering? We argue the answer is yes, under conditions we characterise precisely. We present Act-Observe-Rewrite (AOR), a framework in which an LLM agent improves a robot manipulation policy by synthesising entirely new executable Python controller code between trials, guided by visual observations and structured episode outcomes. Unlike prior work that grounds LLMs in pre-defined skill libraries or uses code generation for one-shot plan synthesis, AOR makes the full low-level motor control implementation the unit of LLM reasoning, enabling the agent to change not just what the robot does, but how it does it. The central claim is that interpretable code as the policy representation creates a qualitatively different kind of in-context learning from opaque neural policies: the agent can diagnose systematic failures and rewrite their causes. We validate this across three robosuite manipulation tasks and report promising results, with the agent achieving high success rates without demonstrations, reward engineering, or gradient updates.
Large Language Models (LLMs) demonstrate potentials for automating scientific code generation but face challenges in reliability, error propagation in multi-agent workflows, and evaluation in domains with ill-defined success metrics. We present a Bayesian adversarial multi-agent framework specifically designed for AI for Science (AI4S) tasks in the form of a Low-code Platform (LCP). Three LLM-based agents are coordinated under the Bayesian framework: a Task Manager that structures user inputs into actionable plans and adaptive test cases, a Code Generator that produces candidate solutions, and an Evaluator providing comprehensive feedback. The framework employs an adversarial loop where the Task Manager iteratively refines test cases to challenge the Code Generator, while prompt distributions are dynamically updated using Bayesian principles by integrating code quality metrics: functional correctness, structural alignment, and static analysis. This co-optimization of tests and code reduces dependence on LLM reliability and addresses evaluation uncertainty inherent to scientific tasks. LCP also streamlines human-AI collaboration by translating non-expert prompts into domain-specific requirements, bypassing the need for manual prompt engineering by practitioners without coding backgrounds. Benchmark evaluations demonstrate LCP's effectiveness in generating robust code while minimizing error propagation. The proposed platform is also tested on an Earth Science cross-disciplinary task and demonstrates strong reliability, outperforming competing models.
In order to flexibly act in an everyday environment, a robotic agent needs a variety of cognitive capabilities that enable it to reason about plans and perform execution recovery. Large language models (LLMs) have been shown to demonstrate emergent cognitive aspects, such as reasoning and language understanding; however, the ability to control embodied robotic agents requires reliably bridging high-level language to low-level functionalities for perception and control. In this paper, we investigate the extent to which an LLM can serve as a core component for planning and execution reasoning in a cognitive robot architecture. For this purpose, we propose a cognitive architecture in which an agentic LLM serves as the core component for planning and reasoning, while components for working and episodic memories support learning from experience and adaptation. An instance of the architecture is then used to control a mobile manipulator in a simulated household environment, where environment interaction is done through a set of high-level tools for perception, reasoning, navigation, grasping, and placement, all of which are made available to the LLM-based agent. We evaluate our proposed system on two household tasks (object placement and object swapping), which evaluate the agent's reasoning, planning, and memory utilisation. The results demonstrate that the LLM-driven agent can complete structured tasks and exhibits emergent adaptation and memory-guided planning, but also reveal significant limitations, such as hallucinations about the task success and poor instruction following by refusing to acknowledge and complete sequential tasks. These findings highlight both the potential and challenges of employing LLMs as embodied cognitive controllers for autonomous robots.
The increasing diversity and scale of video data demand retrieval systems capable of multimodal understanding, adaptive reasoning, and domain-specific knowledge integration. This paper presents LLandMark, a modular multi-agent framework for landmark-aware multimodal video retrieval to handle real-world complex queries. The framework features specialized agents that collaborate across four stages: query parsing and planning, landmark reasoning, multimodal retrieval, and reranked answer synthesis. A key component, the Landmark Knowledge Agent, detects cultural or spatial landmarks and reformulates them into descriptive visual prompts, enhancing CLIP-based semantic matching for Vietnamese scenes. To expand capabilities, we introduce an LLM-assisted image-to-image pipeline, where a large language model (Gemini 2.5 Flash) autonomously detects landmarks, generates image search queries, retrieves representative images, and performs CLIP-based visual similarity matching, removing the need for manual image input. In addition, an OCR refinement module leveraging Gemini and LlamaIndex improves Vietnamese text recognition. Experimental results show that LLandMark achieves adaptive, culturally grounded, and explainable retrieval performance.
Failure is inevitable for embodied navigation in complex environments. To enhance the resilience, replanning (RP) is a viable option, where the robot is allowed to fail, but is capable of adjusting plan until success. However, existing RP approaches freeze the ego action model and miss the opportunities to explore better plans by upgrading the robot itself. To address this limitation, we propose Self-Evolutionary RePlanning, or SERP for short, which leads to a paradigm shift from frozen models towards evolving models by run-time learning from recent experiences. In contrast to existing model evolution approaches that often get stuck at predefined static parameters, we introduce agentic self-evolving action model that uses in-context learning with auto-differentiation (ILAD) for adaptive function adjustment and global parameter reset. To achieve token-efficient replanning for SERP, we also propose graph chain-of-thought (GCOT) replanning with large language model (LLM) inference over distilled graphs. Extensive simulation and real-world experiments demonstrate that SERP achieves higher success rate with lower token expenditure over various benchmarks, validating its superior robustness and efficiency across diverse environments.
While Large Language Models (LLMs) form the cornerstone of sequential decision-making agent development, they have inherent limitations in high-frequency decision tasks. Existing research mainly focuses on discrete embodied decision scenarios with low-frequency and significant semantic differences in state space (e.g., household planning). These methods suffer from limited performance in high-frequency decision-making tasks, since high-precision numerical state information in such tasks undergoes frequent updates with minimal fluctuations, and exhibiting policy misalignment between the learned sub-tasks and composite tasks. To address these issues, this paper proposes Normalized Action Reward guided Consistency Policy Optimization (NAR-CP). 1) Our method first acquires predefined dense rewards from environmental feedback of candidate actions via reward functions, then completes reward shaping through normalization, and theoretically verifies action reward normalization does not impair optimal policy. 2) To reduce policy misalignment in composite tasks, we use LLMs to infer sub-observation candidate actions and generate joint policies, with consistency loss ensuring precise alignment between global semantic policies and sub-semantic policies. Experiments on UAV pursuit, a typical high-frequency task, show our method delivers superior performance on independent and composite tasks with excellent generalization to unseen tasks.
In modern industrial production, multiple robots often collaborate to complete complex manufacturing tasks. Large language models (LLMs), with their strong reasoning capabilities, have shown potential in coordinating robots for simple household and manipulation tasks. However, in industrial scenarios, stricter sequential constraints and more complex dependencies within tasks present new challenges for LLMs. To address this, we propose IMR-LLM, a novel LLM-driven Industrial Multi-Robot task planning and program generation framework. Specifically, we utilize LLMs to assist in constructing disjunctive graphs and employ deterministic solving methods to obtain a feasible and efficient high-level task plan. Based on this, we use a process tree to guide LLMs to generate executable low-level programs. Additionally, we create IMR-Bench, a challenging benchmark that encompasses multi-robot industrial tasks across three levels of complexity. Experimental results indicate that our method significantly surpasses existing methods across all evaluation metrics.
Autonomous driving systems require comprehensive evaluation in safety-critical scenarios to ensure safety and robustness. However, such scenarios are rare and difficult to collect from real-world driving data, necessitating simulation-based synthesis. Yet, existing methods often exhibit limitations in both controllability and realism. From a capability perspective, LLMs excel at controllable generation guided by natural language instructions, while diffusion models are better suited for producing trajectories consistent with realistic driving distributions. Leveraging their complementary strengths, we propose AnchorDrive, a two-stage safety-critical scenario generation framework. In the first stage, we deploy an LLM as a driver agent within a closed-loop simulation, which reasons and iteratively outputs control commands under natural language constraints; a plan assessor reviews these commands and provides corrective feedback, enabling semantically controllable scenario generation. In the second stage, the LLM extracts key anchor points from the first-stage trajectories as guidance objectives, which jointly with other guidance terms steer the diffusion model to regenerate complete trajectories with improved realism while preserving user-specified intent. Experiments on the highD dataset demonstrate that AnchorDrive achieves superior overall performance in criticality, realism, and controllability, validating its effectiveness for generating controllable and realistic safety-critical scenarios.
Large language models (LLMs) are increasingly applied to health management, showing promise across disease prevention, clinical decision-making, and long-term care. However, existing medical benchmarks remain largely static and task-isolated, failing to capture the openness, longitudinal structure, and safety-critical complexity of real-world clinical workflows. We introduce ClinConsensus, a Chinese medical benchmark curated, validated, and quality-controlled by clinical experts. ClinConsensus comprises 2500 open-ended cases spanning the full continuum of care--from prevention and intervention to long-term follow-up--covering 36 medical specialties, 12 common clinical task types, and progressively increasing levels of complexity. To enable reliable evaluation of such complex scenarios, we adopt a rubric-based grading protocol and propose the Clinically Applicable Consistency Score (CACS@k). We further introduce a dual-judge evaluation framework, combining a high-capability LLM-as-judge with a distilled, locally deployable judge model trained via supervised fine-tuning, enabling scalable and reproducible evaluation aligned with physician judgment. Using ClinConsensus, we conduct a comprehensive assessment of several leading LLMs and reveal substantial heterogeneity across task themes, care stages, and medical specialties. While top-performing models achieve comparable overall scores, they differ markedly in reasoning, evidence use, and longitudinal follow-up capabilities, and clinically actionable treatment planning remains a key bottleneck. We release ClinConsensus as an extensible benchmark to support the development and evaluation of medical LLMs that are robust, clinically grounded, and ready for real-world deployment.
When automating plan generation for a real-world sequential decision problem, the goal is often not to replace the human planner, but to facilitate an iterative reasoning and elicitation process, where the human's role is to guide the AI planner according to their preferences and expertise. In this context, explanations that respond to users' questions are crucial to improve their understanding of potential solutions and increase their trust in the system. To enable natural interaction with such a system, we present a multi-agent Large Language Model (LLM) architecture that is agnostic to the explanation framework and enables user- and context-dependent interactive explanations. We also describe an instantiation of this framework for goal-conflict explanations, which we use to conduct a user study comparing the LLM-powered interaction with a baseline template-based explanation interface.
Reinforcement learning (RL) has demonstrated notable success in post-training large language models (LLMs) as agents for tasks such as computer use, tool calling, and coding. However, exploration remains a central challenge in RL for LLM agents, especially as they operate in language-action spaces with complex observations and sparse outcome rewards. In this work, we address exploration for LLM agents by leveraging the ability of LLMs to plan and reason in language about the environment to shift exploration from low-level actions to higher-level language strategies. We thus propose Strategy-Guided Exploration (SGE), which first generates a concise natural-language strategy that describes what to do to make progress toward the goal, and then generates environment actions conditioned on that strategy. By exploring in the space of strategies rather than the space of actions, SGE induces structured and diverse exploration that targets different environment outcomes. To increase strategy diversity during RL, SGE introduces mixed-temperature sampling, which explores diverse strategies in parallel, along with a strategy reflection process that grounds strategy generation on the outcomes of previous strategies in the environment. Across UI interaction, tool-calling, coding, and embodied agent environments, SGE consistently outperforms exploration-focused RL baselines, improving both learning efficiency and final performance. We show that SGE enables the agent to learn to solve tasks too difficult for the base model.
Interactive articles help readers engage with complex ideas through exploration, yet creating them remains costly, requiring both domain expertise and web development skills. Recent LLM-based agents can automate content creation, but naively applying them yields uncontrollable and unverifiable outputs. We present ViviDoc, a human-agent collaborative system that generates interactive educational documents from a single topic input. ViviDoc introduces a multi-agent pipeline (Planner, Executor, Evaluator) and the Document Specification (DocSpec), a human-readable intermediate representation that decomposes each interactive visualization into State, Render, Transition, and Constraint components. The DocSpec enables educators to review and refine generation plans before code is produced, bridging the gap between pedagogical intent and executable output. Expert evaluation and a user study show that ViviDoc substantially outperforms naive agentic generation and provides an intuitive editing experience. Our project homepage is available at https://vividoc-homepage.vercel.app/.
Large Language Models (LLMs) are increasingly deployed as agentic systems that plan, memorize, and act in open-world environments. This shift brings new security problems: failures are no longer only unsafe text generation, but can become real harm through tool use, persistent memory, and interaction with untrusted web content. In this survey, we provide a transition-oriented view from Secure Agentic AI to a Secure Agentic Web. We first summarize a component-aligned threat taxonomy covering prompt abuse, environment injection, memory attacks, toolchain abuse, model tampering, and agent network attacks. We then review defense strategies, including prompt hardening, safety-aware decoding, privilege control for tools and APIs, runtime monitoring, continuous red-teaming, and protocol-level security mechanisms. We further discuss how these threats and mitigations escalate in the Agentic Web, where delegation chains, cross-domain interactions, and protocol-mediated ecosystems amplify risks via propagation and composition. Finally, we highlight open challenges for web-scale deployment, such as interoperable identity and authorization, provenance and traceability, ecosystem-level response, and scalable evaluation under adaptive adversaries. Our goal is to connect recent empirical findings with system-level requirements, and to outline practical research directions toward trustworthy agent ecosystems.
Tool-using LLM agents face a reliability-cost tradeoff: routing every decision through the LLM improves correctness but incurs high latency and inference cost, while pre-coded workflow graphs reduce cost but become brittle under unanticipated compound tool failures. We present Self-Healing Router, a fault-tolerant orchestration architecture that treats most agent control-flow decisions as routing rather than reasoning. The system combines (i) parallel health monitors that assign priority scores to runtime conditions such as tool outages and risk signals, and (ii) a cost-weighted tool graph where Dijkstra's algorithm performs deterministic shortest-path routing. When a tool fails mid-execution, its edges are reweighted to infinity and the path is recomputed -- yielding automatic recovery without invoking the LLM. The LLM is reserved exclusively for cases where no feasible path exists, enabling goal demotion or escalation. Prior graph-based tool-use systems (ControlLLM, ToolNet, NaviAgent) focus on tool selection and planning; our contribution is runtime fault tolerance with deterministic recovery and binary observability -- every failure is either a logged reroute or an explicit escalation, never a silent skip. Across 19 scenarios spanning three graph topologies (linear pipeline, dependency DAG, parallel fan-out), Self-Healing Router matches ReAct's correctness while reducing control-plane LLM calls by 93% (9 vs 123 aggregate) and eliminating the silent-failure cases observed in a well-engineered static workflow baseline under compound failures.