Recent advancements in Large Language Models (LLMs) and their multimodal extensions (MLLMs) have substantially enhanced machine reasoning across diverse tasks. However, these models predominantly rely on pure text as the medium for both expressing and structuring reasoning, even when visual information is present. In this work, we argue that language may not always be the most natural or effective modality for reasoning, particularly in tasks involving spatial and geometrical information. Motivated by this, we propose a new paradigm, Visual Planning, which enables planning through purely visual representations, independent of text. In this paradigm, planning is executed via sequences of images that encode step-by-step inference in the visual domain, akin to how humans sketch or visualize future actions. We introduce a novel reinforcement learning framework, Visual Planning via Reinforcement Learning (VPRL), empowered by GRPO for post-training large vision models, leading to substantial improvements in planning in a selection of representative visual navigation tasks, FrozenLake, Maze, and MiniBehavior. Our visual planning paradigm outperforms all other planning variants that conduct reasoning in the text-only space. Our results establish Visual Planning as a viable and promising alternative to language-based reasoning, opening new avenues for tasks that benefit from intuitive, image-based inference.
Motivated by Smart Manufacturing and Industry 4.0, we introduce a framework for synthesizing Abstraction-Based Controller Design (ABCD) for reach-avoid problems from Natural Language (NL) specifications using Large Language Models (LLMs). A Code Agent interprets an NL description of the control problem and translates it into a formal language interpretable by state-of-the-art symbolic control software, while a Checker Agent verifies the correctness of the generated code and enhances safety by identifying specification mismatches. Evaluations show that the system handles linguistic variability and improves robustness over direct planning with LLMs. The proposed approach lowers the barrier to formal control synthesis by enabling intuitive, NL-based task definition while maintaining safety guarantees through automated validation.
RAG can enhance the performance of LLMs on knowledge-intensive tasks. Various RAG paradigms, including vanilla, planning-based, and iterative RAG, are built upon 2 cores: the retriever, which should robustly select relevant documents across complex queries, and the generator, which should faithfully synthesize responses. However, existing retrievers rely heavily on public knowledge and struggle with queries of varying logical complexity and clue completeness, while generators frequently face fidelity problems. In this work, we introduce RAGSynth, a framework that includes a data construction modeling and a corresponding synthetic data generation implementation, designed to optimize retriever robustness and generator fidelity. Additionally, we present SynthBench, a benchmark encompassing 8 domain-specific documents across 4 domains, featuring diverse query complexities, clue completeness, and fine-grained citation granularity. Leveraging RAGSynth, we generate a large-scale synthetic dataset, including single and multi-hop. Extensive experiments demonstrate that the synthetic data significantly improves the robustness of the retrievers and the fidelity of the generators. Additional evaluations confirm that RAGSynth can also generalize well across different domains. By integrating the optimized retrievers into various RAG paradigms, we consistently observe enhanced RAG system performance. We have open-sourced the implementation on https://github.com/EachSheep/RAGSynth.
Spatiotemporal relationships are critical in data science, as many prediction and reasoning tasks require analysis across both spatial and temporal dimensions--for instance, navigating an unfamiliar city involves planning itineraries that sequence locations and timing cultural experiences. However, existing Question-Answering (QA) datasets lack sufficient spatiotemporal-sensitive questions, making them inadequate benchmarks for evaluating models' spatiotemporal reasoning capabilities. To address this gap, we introduce POI-QA, a novel spatiotemporal-sensitive QA dataset centered on Point of Interest (POI), constructed through three key steps: mining and aligning open-source vehicle trajectory data from GAIA with high-precision geographic POI data, rigorous manual validation of noisy spatiotemporal facts, and generating bilingual (Chinese/English) QA pairs that reflect human-understandable spatiotemporal reasoning tasks. Our dataset challenges models to parse complex spatiotemporal dependencies, and evaluations of state-of-the-art multilingual LLMs (e.g., Qwen2.5-7B, Llama3.1-8B) reveal stark limitations: even the top-performing model (Qwen2.5-7B fine-tuned with RAG+LoRA) achieves a top 10 Hit Ratio (HR@10) of only 0.41 on the easiest task, far below human performance at 0.56. This underscores persistent weaknesses in LLMs' ability to perform consistent spatiotemporal reasoning, while highlighting POI-QA as a robust benchmark to advance algorithms sensitive to spatiotemporal dynamics. The dataset is publicly available at https://www.kaggle.com/ds/7394666.
Planning trips is a cognitively intensive task involving conflicting user preferences, dynamic external information, and multi-step temporal-spatial optimization. Traditional platforms often fall short - they provide static results, lack contextual adaptation, and fail to support real-time interaction or intent refinement. Our approach, Vaiage, addresses these challenges through a graph-structured multi-agent framework built around large language models (LLMs) that serve as both goal-conditioned recommenders and sequential planners. LLMs infer user intent, suggest personalized destinations and activities, and synthesize itineraries that align with contextual constraints such as budget, timing, group size, and weather. Through natural language interaction, structured tool use, and map-based feedback loops, Vaiage enables adaptive, explainable, and end-to-end travel planning grounded in both symbolic reasoning and conversational understanding. To evaluate Vaiage, we conducted human-in-the-loop experiments using rubric-based GPT-4 assessments and qualitative feedback. The full system achieved an average score of 8.5 out of 10, outperforming the no-strategy (7.2) and no-external-API (6.8) variants, particularly in feasibility. Qualitative analysis indicated that agent coordination - especially the Strategy and Information Agents - significantly improved itinerary quality by optimizing time use and integrating real-time context. These results demonstrate the effectiveness of combining LLM reasoning with symbolic agent coordination in open-ended, real-world planning tasks.
Robot task planning decomposes human instructions into executable action sequences that enable robots to complete a series of complex tasks. Although recent large language model (LLM)-based task planners achieve amazing performance, they assume that human instructions are clear and straightforward. However, real-world users are not experts, and their instructions to robots often contain significant vagueness. Linguists suggest that such vagueness frequently arises from referring expressions (REs), whose meanings depend heavily on dialogue context and environment. This vagueness is even more prevalent among the elderly and children, who robots should serve more. This paper studies how such vagueness in REs within human instructions affects LLM-based robot task planning and how to overcome this issue. To this end, we propose the first robot task planning benchmark with vague REs (REI-Bench), where we discover that the vagueness of REs can severely degrade robot planning performance, leading to success rate drops of up to 77.9%. We also observe that most failure cases stem from missing objects in planners. To mitigate the REs issue, we propose a simple yet effective approach: task-oriented context cognition, which generates clear instructions for robots, achieving state-of-the-art performance compared to aware prompt and chains of thought. This work contributes to the research community of human-robot interaction (HRI) by making robot task planning more practical, particularly for non-expert users, e.g., the elderly and children.
Learning how the world works is central to building AI agents that can adapt to complex environments. Traditional world models based on deep learning demand vast amounts of training data, and do not flexibly update their knowledge from sparse observations. Recent advances in program synthesis using Large Language Models (LLMs) give an alternate approach which learns world models represented as source code, supporting strong generalization from little data. To date, application of program-structured world models remains limited to natural language and grid-world domains. We introduce a novel program synthesis method for effectively modeling complex, non-gridworld domains by representing a world model as an exponentially-weighted product of programmatic experts (PoE-World) synthesized by LLMs. We show that this approach can learn complex, stochastic world models from just a few observations. We evaluate the learned world models by embedding them in a model-based planning agent, demonstrating efficient performance and generalization to unseen levels on Atari's Pong and Montezuma's Revenge. We release our code and display the learned world models and videos of the agent's gameplay at https://topwasu.github.io/poe-world.
We propose an iterative programmatic planning (IPP) framework for solving grid-based tasks by synthesizing interpretable agent policies expressed in code using large language models (LLMs). Instead of relying on traditional search or reinforcement learning, our approach uses code generation as policy synthesis, where the LLM outputs executable programs that map environment states to action sequences. Our proposed architecture incorporates several prompting strategies, including direct code generation, pseudocode-conditioned refinement, and curriculum-based prompting, but also includes an iterative refinement mechanism that updates code based on task performance feedback. We evaluate our approach using six leading LLMs and two challenging grid-based benchmarks (GRASP and MiniGrid). Our IPP framework demonstrates improvements over direct code generation ranging from 10\% to as much as 10x across five of the six models and establishes a new state-of-the-art result for GRASP. IPP is found to significantly outperform direct elicitation of a solution from GPT-o3-mini (by 63\% on MiniGrid to 116\% on GRASP), demonstrating the viability of the overall approach. Computational costs of all code generation approaches are similar. While code generation has a higher initial prompting cost compared to direct solution elicitation (\$0.08 per task vs. \$0.002 per instance for GPT-o3-mini), the code can be reused for any number of instances, making the amortized cost significantly lower (by 400x on GPT-o3-mini across the complete GRASP benchmark).
Responsible AI (RAI) efforts increasingly emphasize the importance of addressing potential harms early in the AI development lifecycle through social-technical lenses. However, in cross-functional industry teams, this work is often stalled by a persistent knowledge handoff challenge: the difficulty of transferring high-level, early-stage technical design rationales from technical experts to non-technical or user-facing roles for ethical evaluation and harm identification. Through literature review and a co-design study with 8 practitioners, we unpack how this challenge manifests -- technical design choices are rarely handed off in ways that support meaningful engagement by non-technical roles; collaborative workflows lack shared, visual structures to support mutual understanding; and non-technical practitioners are left without scaffolds for systematic harm evaluation. Existing tools like JIRA or Google Docs, while useful for product tracking, are ill-suited for supporting joint harm identification across roles, often requiring significant extra effort to align understanding. To address this, we developed AI LEGO, a web-based prototype that supports cross-functional AI practitioners in effectively facilitating knowledge handoff and identifying harmful design choices in the early design stages. Technical roles use interactive blocks to draft development plans, while non-technical roles engage with those blocks through stage-specific checklists and LLM-driven persona simulations to surface potential harms. In a study with 18 cross-functional practitioners, AI LEGO increased the volume and likelihood of harms identified compared to baseline worksheets. Participants found that its modular structure and persona prompts made harm identification more accessible, fostering clearer and more collaborative RAI practices in early design.
The ReAct (Reasoning + Action) capability in large language models (LLMs) has become the foundation of modern agentic systems. Recent LLMs, such as DeepSeek-R1 and OpenAI o1/o3, exemplify this by emphasizing reasoning through the generation of ample intermediate tokens, which help build a strong premise before producing the final output tokens. In this paper, we introduce Pre-Act, a novel approach that enhances the agent's performance by creating a multi-step execution plan along with the detailed reasoning for the given user input. This plan incrementally incorporates previous steps and tool outputs, refining itself after each step execution until the final response is obtained. Our approach is applicable to both conversational and non-conversational agents. To measure the performance of task-oriented agents comprehensively, we propose a two-level evaluation framework: (1) turn level and (2) end-to-end. Our turn-level evaluation, averaged across five models, shows that our approach, Pre-Act, outperforms ReAct by 70% in Action Recall on the Almita dataset. While this approach is effective for larger models, smaller models crucial for practical applications, where latency and cost are key constraints, often struggle with complex reasoning tasks required for agentic systems. To address this limitation, we fine-tune relatively small models such as Llama 3.1 (8B & 70B) using the proposed Pre-Act approach. Our experiments show that the fine-tuned 70B model outperforms GPT-4, achieving a 69.5% improvement in action accuracy (turn-level) and a 28% improvement in goal completion rate (end-to-end) on the Almita (out-of-domain) dataset.
The proliferation of Large Language Model (LLM)-based Graphical User Interface (GUI) agents in web browsing scenarios present complex unintended consequences (UCs). This paper characterizes three UCs from three perspectives: phenomena, influence and mitigation, drawing on social media analysis (N=221 posts) and semi-structured interviews (N=14). Key phenomenon for UCs include agents' deficiencies in comprehending instructions and planning tasks, challenges in executing accurate GUI interactions and adapting to dynamic interfaces, the generation of unreliable or misaligned outputs, and shortcomings in error handling and feedback processing. These phenomena manifest as influences from unanticipated actions and user frustration, to privacy violations and security vulnerabilities, and further to eroded trust and wider ethical concerns. Our analysis also identifies user-initiated mitigation, such as technical adjustments and manual oversight, and provides implications for designing future LLM-based GUI agents that are robust, user-centric, and transparent, fostering a crucial balance between automation and human oversight.
Large Language Models (LLMs) have shown impressive performance across natural language tasks, but their ability to forecast violent conflict remains underexplored. We investigate whether LLMs possess meaningful parametric knowledge-encoded in their pretrained weights-to predict conflict escalation and fatalities without external data. This is critical for early warning systems, humanitarian planning, and policy-making. We compare this parametric knowledge with non-parametric capabilities, where LLMs access structured and unstructured context from conflict datasets (e.g., ACLED, GDELT) and recent news reports via Retrieval-Augmented Generation (RAG). Incorporating external information could enhance model performance by providing up-to-date context otherwise missing from pretrained weights. Our two-part evaluation framework spans 2020-2024 across conflict-prone regions in the Horn of Africa and the Middle East. In the parametric setting, LLMs predict conflict trends and fatalities relying only on pretrained knowledge. In the non-parametric setting, models receive summaries of recent conflict events, indicators, and geopolitical developments. We compare predicted conflict trend labels (e.g., Escalate, Stable Conflict, De-escalate, Peace) and fatalities against historical data. Our findings highlight the strengths and limitations of LLMs for conflict forecasting and the benefits of augmenting them with structured external knowledge.
The integration of foundation models (FMs) into robotics has enabled robots to understand natural language and reason about the semantics in their environments. However, existing FM-enabled robots primary operate in closed-world settings, where the robot is given a full prior map or has a full view of its workspace. This paper addresses the deployment of FM-enabled robots in the field, where missions often require a robot to operate in large-scale and unstructured environments. To effectively accomplish these missions, robots must actively explore their environments, navigate obstacle-cluttered terrain, handle unexpected sensor inputs, and operate with compute constraints. We discuss recent deployments of SPINE, our LLM-enabled autonomy framework, in field robotic settings. To the best of our knowledge, we present the first demonstration of large-scale LLM-enabled robot planning in unstructured environments with several kilometers of missions. SPINE is agnostic to a particular LLM, which allows us to distill small language models capable of running onboard size, weight and power (SWaP) limited platforms. Via preliminary model distillation work, we then present the first language-driven UAV planner using on-device language models. We conclude our paper by proposing several promising directions for future research.
Large Language Models (LLMs) show growing promise in autonomous driving by reasoning over complex traffic scenarios to generate path plans. However, their tendencies toward overconfidence, and hallucinations raise critical safety concerns. We introduce SafePath, a modular framework that augments LLM-based path planning with formal safety guarantees using conformal prediction. SafePath operates in three stages. In the first stage, we use an LLM that generates a set of diverse candidate paths, exploring possible trajectories based on agent behaviors and environmental cues. In the second stage, SafePath filters out high-risk trajectories while guaranteeing that at least one safe option is included with a user-defined probability, through a multiple-choice question-answering formulation that integrates conformal prediction. In the final stage, our approach selects the path with the lowest expected collision risk when uncertainty is low or delegates control to a human when uncertainty is high. We theoretically prove that SafePath guarantees a safe trajectory with a user-defined probability, and we show how its human delegation rate can be tuned to balance autonomy and safety. Extensive experiments on nuScenes and Highway-env show that SafePath reduces planning uncertainty by 77\% and collision rates by up to 70\%, demonstrating effectiveness in making LLM-driven path planning more safer.
Generative Artificial Intelligence (GenAI) constitutes a transformative technological wave that reconfigures industries through its unparalleled capabilities for content creation, reasoning, planning, and multimodal understanding. This revolutionary force offers the most promising path yet toward solving one of engineering's grandest challenges: achieving reliable, fully autonomous driving, particularly the pursuit of Level 5 autonomy. This survey delivers a comprehensive and critical synthesis of the emerging role of GenAI across the autonomous driving stack. We begin by distilling the principles and trade-offs of modern generative modeling, encompassing VAEs, GANs, Diffusion Models, and Large Language Models (LLMs). We then map their frontier applications in image, LiDAR, trajectory, occupancy, video generation as well as LLM-guided reasoning and decision making. We categorize practical applications, such as synthetic data workflows, end-to-end driving strategies, high-fidelity digital twin systems, smart transportation networks, and cross-domain transfer to embodied AI. We identify key obstacles and possibilities such as comprehensive generalization across rare cases, evaluation and safety checks, budget-limited implementation, regulatory compliance, ethical concerns, and environmental effects, while proposing research plans across theoretical assurances, trust metrics, transport integration, and socio-technical influence. By unifying these threads, the survey provides a forward-looking reference for researchers, engineers, and policymakers navigating the convergence of generative AI and advanced autonomous mobility. An actively maintained repository of cited works is available at https://github.com/taco-group/GenAI4AD.
Generative AI can help small business owners automate tasks, increase efficiency, and improve their bottom line. However, despite the seemingly intuitive design of systems like ChatGPT, significant barriers remain for those less comfortable with technology. To address these disparities, prior work highlights accessory skills -- beyond prompt engineering -- users must master to successfully adopt generative AI including keyboard shortcuts, editing skills, file conversions, and browser literacy. Building on a design workshop series and 15 interviews with small businesses, we introduce BizChat, a large language model (LLM)-powered web application that helps business owners across digital skills levels write their business plan -- an essential but often neglected document. To do so, BizChat's interface embodies three design considerations inspired by learning sciences: ensuring accessibility to users with less digital skills while maintaining extensibility to power users ("low-floor-high-ceiling"), providing in situ micro-learning to support entrepreneurial education ("just-in-time learning"), and framing interaction around business activities ("contextualized technology introduction"). We conclude with plans for a future BizChat deployment.
PDDL-based symbolic task planning remains pivotal for robot autonomy yet struggles with dynamic human-robot collaboration due to scalability, re-planning demands, and delayed plan availability. Although a few neurosymbolic frameworks have previously leveraged LLMs such as GPT-3 to address these challenges, reliance on closed-source, remote models with limited context introduced critical constraints: third-party dependency, inconsistent response times, restricted plan length and complexity, and multi-domain scalability issues. We present Gideon, a novel framework that enables the transition to modern, smaller, local LLMs with extended context length. Gideon integrates a novel problem generator to systematically generate large-scale datasets of realistic domain-problem-plan tuples for any domain, and adapts neurosymbolic planning for local LLMs, enabling on-device execution and extended context for multi-domain support. Preliminary experiments in single-domain scenarios performed on Qwen-2.5 1.5B and trained on 8k-32k samples, demonstrate a valid plan percentage of 66.1% (32k model) and show that the figure can be further scaled through additional data. Multi-domain tests on 16k samples yield an even higher 70.6% planning validity rate, proving extensibility across domains and signaling that data variety can have a positive effect on learning efficiency. Although long-horizon planning and reduced model size make Gideon training much less efficient than baseline models based on larger LLMs, the results are still significant considering that the trained model is about 120x smaller than baseline and that significant advantages can be achieved in inference efficiency, scalability, and multi-domain adaptability, all critical factors in human-robot collaboration. Training inefficiency can be mitigated by Gideon's streamlined data generation pipeline.
Efficiently modeling and exploiting opponents is a long-standing challenge in adversarial domains. Large Language Models (LLMs) trained on extensive textual data have recently demonstrated outstanding performance in general tasks, introducing new research directions for opponent modeling. Some studies primarily focus on directly using LLMs to generate decisions based on the elaborate prompt context that incorporates opponent descriptions, while these approaches are limited to scenarios where LLMs possess adequate domain expertise. To address that, we introduce a two-stage Strategy-Augmented Planning (SAP) framework that significantly enhances the opponent exploitation capabilities of LLM-based agents by utilizing a critical component, the Strategy Evaluation Network (SEN). Specifically, in the offline stage, we construct an explicit strategy space and subsequently collect strategy-outcome pair data for training the SEN network. During the online phase, SAP dynamically recognizes the opponent's strategies and greedily exploits them by searching best response strategy on the well-trained SEN, finally translating strategy to a course of actions by carefully designed prompts. Experimental results show that SAP exhibits robust generalization capabilities, allowing it to perform effectively not only against previously encountered opponent strategies but also against novel, unseen strategies. In the MicroRTS environment, SAP achieves a 85.35\% performance improvement over baseline methods and matches the competitiveness of reinforcement learning approaches against state-of-the-art (SOTA) rule-based AI.
The remote embodied referring expression (REVERIE) task requires an agent to navigate through complex indoor environments and localize a remote object specified by high-level instructions, such as "bring me a spoon", without pre-exploration. Hence, an efficient navigation plan is essential for the final success. This paper proposes a novel parameter-efficient action planner using large language models (PEAP-LLM) to generate a single-step instruction at each location. The proposed model consists of two modules, LLM goal planner (LGP) and LoRA action planner (LAP). Initially, LGP extracts the goal-oriented plan from REVERIE instructions, including the target object and room. Then, LAP generates a single-step instruction with the goal-oriented plan, high-level instruction, and current visual observation as input. PEAP-LLM enables the embodied agent to interact with LAP as the path planner on the fly. A simple direct application of LLMs hardly achieves good performance. Also, existing hard-prompt-based methods are error-prone in complicated scenarios and need human intervention. To address these issues and prevent the LLM from generating hallucinations and biased information, we propose a novel two-stage method for fine-tuning the LLM, consisting of supervised fine-tuning (STF) and direct preference optimization (DPO). SFT improves the quality of generated instructions, while DPO utilizes environmental feedback. Experimental results show the superiority of our proposed model on REVERIE compared to the previous state-of-the-art.
We present UAV-CodeAgents, a scalable multi-agent framework for autonomous UAV mission generation, built on large language and vision-language models (LLMs/VLMs). The system leverages the ReAct (Reason + Act) paradigm to interpret satellite imagery, ground high-level natural language instructions, and collaboratively generate UAV trajectories with minimal human supervision. A core component is a vision-grounded, pixel-pointing mechanism that enables precise localization of semantic targets on aerial maps. To support real-time adaptability, we introduce a reactive thinking loop, allowing agents to iteratively reflect on observations, revise mission goals, and coordinate dynamically in evolving environments. UAV-CodeAgents is evaluated on large-scale mission scenarios involving industrial and environmental fire detection. Our results show that a lower decoding temperature (0.5) yields higher planning reliability and reduced execution time, with an average mission creation time of 96.96 seconds and a success rate of 93%. We further fine-tune Qwen2.5VL-7B on 9,000 annotated satellite images, achieving strong spatial grounding across diverse visual categories. To foster reproducibility and future research, we will release the full codebase and a novel benchmark dataset for vision-language-based UAV planning.
Automated penetration testing (AutoPT) powered by large language models (LLMs) has gained attention for its ability to automate ethical hacking processes and identify vulnerabilities in target systems by leveraging the intrinsic knowledge of LLMs. However, existing LLM-based AutoPT frameworks often underperform compared to human experts in challenging tasks for several reasons: the imbalanced knowledge used in LLM training, short-sighted planning in the planning process, and hallucinations during command generation. In addition, the penetration testing (PT) process, with its trial-and-error nature, is limited by existing frameworks that lack mechanisms to learn from previous failed operations, restricting adaptive improvement of PT strategies. To address these limitations, we propose a knowledge-informed self-reflective PT framework powered by LLMs, called RefPentester, which is an AutoPT framework designed to assist human operators in identifying the current stage of the PT process, selecting appropriate tactic and technique for the stage, choosing suggested action, providing step-by-step operational guidance, and learning from previous failed operations. We also modeled the PT process as a seven-state Stage Machine to integrate the proposed framework effectively. The evaluation shows that RefPentester can successfully reveal credentials on Hack The Box's Sau machine, outperforming the baseline GPT-4o model by 16.7%. Across PT stages, RefPentester also demonstrates superior success rates on PT stage transitions.
Retrosynthesis, the process of breaking down a target molecule into simpler precursors through a series of valid reactions, stands at the core of organic chemistry and drug development. Although recent machine learning (ML) research has advanced single-step retrosynthetic modeling and subsequent route searches, these solutions remain restricted by the extensive combinatorial space of possible pathways. Concurrently, large language models (LLMs) have exhibited remarkable chemical knowledge, hinting at their potential to tackle complex decision-making tasks in chemistry. In this work, we explore whether LLMs can successfully navigate the highly constrained, multi-step retrosynthesis planning problem. We introduce an efficient scheme for encoding reaction pathways and present a new route-level search strategy, moving beyond the conventional step-by-step reactant prediction. Through comprehensive evaluations, we show that our LLM-augmented approach excels at retrosynthesis planning and extends naturally to the broader challenge of synthesizable molecular design.
Emotional support conversation (ESC) aims to alleviate the emotional distress of individuals through effective conversations. Although large language models (LLMs) have obtained remarkable progress on ESC, most of these studies might not define the diagram from the state model perspective, therefore providing a suboptimal solution for long-term satisfaction. To address such an issue, we leverage the Q-learning on LLMs, and propose a framework called straQ*. Our framework allows a plug-and-play LLM to bootstrap the planning during ESC, determine the optimal strategy based on long-term returns, and finally guide the LLM to response. Substantial experiments on ESC datasets suggest that straQ* outperforms many baselines, including direct inference, self-refine, chain of thought, finetuning, and finite state machines.
With the widespread application of large AI models in various fields, the automation level of multi-agent systems has been continuously improved. However, in high-risk decision-making scenarios such as healthcare and finance, human participation and the alignment of intelligent systems with human intentions remain crucial. This paper focuses on the financial scenario and constructs a multi-agent brainstorming framework based on the BDI theory. A human-computer collaborative multi-agent financial analysis process is built using Streamlit. The system plans tasks according to user intentions, reduces users' cognitive load through real-time updated structured text summaries and the interactive Cothinker module, and reasonably integrates general and reasoning large models to enhance the ability to handle complex problems. By designing a quantitative analysis algorithm for the sentiment tendency of interview content based on LLMs and a method for evaluating the diversity of ideas generated by LLMs in brainstorming based on k-means clustering and information entropy, the system is comprehensively evaluated. The results of human factors testing show that the system performs well in terms of usability and user experience. Although there is still room for improvement, it can effectively support users in completing complex financial tasks. The research shows that the system significantly improves the efficiency of human-computer interaction and the quality of decision-making in financial decision-making scenarios, providing a new direction for the development of related fields.
Current hardware security verification processes predominantly rely on manual threat modeling and test plan generation, which are labor-intensive, error-prone, and struggle to scale with increasing design complexity and evolving attack methodologies. To address these challenges, we propose ThreatLens, an LLM-driven multi-agent framework that automates security threat modeling and test plan generation for hardware security verification. ThreatLens integrates retrieval-augmented generation (RAG) to extract relevant security knowledge, LLM-powered reasoning for threat assessment, and interactive user feedback to ensure the generation of practical test plans. By automating these processes, the framework reduces the manual verification effort, enhances coverage, and ensures a structured, adaptable approach to security verification. We evaluated our framework on the NEORV32 SoC, demonstrating its capability to automate security verification through structured test plans and validating its effectiveness in real-world scenarios.
Due to time constraints, mental health professionals in China are unable to offer patients prolonged talk therapy, leaving a gap in care for patients with psychological disorders, including aberrant sleep and eating patterns, maladaptive explanatory styles, and gastrointestinal dysfunction. To bridge this gap in care and address these problems in a large-scale manner, we built NeuroPal, a large language model (LLM)-assistant that provides scalable, evidence-based interventions with three clinically validated modules: (1) a sleep chronotherapy planner to output personalized circadian rhythm correction protocols, (2) a cognitive-behavioral reframing engine grounded in CBT and humanistic principles to shift negative attributional biases, and (3) a biochemical regulation advisor to output phytotherapy formulations to regulate sleep-metabolism-gut-axis imbalances. In collaboration with Peking Union Medical College Hospital and Xiangya Hospital Central South University, we ran an RCT protocol with 513 participants with mood/anxiety disorders and showed statistically significant improvements towards primary endpoints (> p<.01). Experiment shows 37.2% drop in the Pittsburgh Sleep Quality Index (PSQI), 28.6% rise in positive affective word usages (LIWC analysis), and 23.4% improvement in patient-reported digestive comfort. The assistant also reached 89.1% adherence rates, significantly higher than the human-guided therapy intervention (72.3%) in matched controls. Our results indicate that an LLM-driven multimodal intervention is able to successfully bridge time-constrained clinical practice while preserving therapeutic effectiveness. For next steps, we plan to explore longitudinal outcome tracking and FDA/CFDA certification routes.
Large Language Models (LLMs) have advanced rapidly in recent years, demonstrating strong capabilities in problem comprehension and reasoning. Inspired by these developments, researchers have begun exploring the use of LLMs as decentralized decision-makers for multi-robot formation control. However, prior studies reveal that directly applying LLMs to such tasks often leads to unstable and inconsistent behaviors, where robots may collapse to the centroid of their positions or diverge entirely due to hallucinated reasoning, logical inconsistencies, and limited coordination awareness. To overcome these limitations, we propose a novel framework that integrates LLMs with an influence-based plan consensus protocol. In this framework, each robot independently generates a local plan toward the desired formation using its own LLM. The robots then iteratively refine their plans through a decentralized consensus protocol that accounts for their influence on neighboring robots. This process drives the system toward a coherent and stable flocking formation in a fully decentralized manner. We evaluate our approach through comprehensive simulations involving both state-of-the-art closed-source LLMs (e.g., o3-mini, Claude 3.5) and open-source models (e.g., Llama3.1-405b, Qwen-Max, DeepSeek-R1). The results show notable improvements in stability, convergence, and adaptability over previous LLM-based methods. We further validate our framework on a physical team of Crazyflie drones, demonstrating its practical viability and effectiveness in real-world multi-robot systems.
Retouching is an essential task in post-manipulation of raw photographs. Generative editing, guided by text or strokes, provides a new tool accessible to users but can easily change the identity of the original objects in unacceptable and unpredictable ways. In contrast, although traditional procedural edits, as commonly supported by photoediting tools (e.g., Gimp, Lightroom), are conservative, they are still preferred by professionals. Unfortunately, professional quality retouching involves many individual procedural editing operations that is challenging to plan for most novices. In this paper, we ask if a multimodal large language model (MLLM) can be taught to critique raw photographs, suggest suitable remedies, and finally realize them with a given set of pre-authored procedural image operations. We demonstrate that MLLMs can be first made aware of the underlying image processing operations, by training them to solve specially designed visual puzzles. Subsequently, such an operation-aware MLLM can both plan and propose edit sequences. To facilitate training, given a set of expert-edited photos, we synthesize a reasoning dataset by procedurally manipulating the expert edits and then grounding a pretrained LLM on the visual adjustments, to synthesize reasoning for finetuning. The proposed retouching operations are, by construction, understandable by the users, preserve object details and resolution, and can be optionally overridden. We evaluate our setup on a variety of test examples and show advantages, in terms of explainability and identity preservation, over existing generative and other procedural alternatives. Code, data, models, and supplementary results can be found via our project website at https://monetgpt.github.io.
We introduce ELA-ZSON, an efficient layout-aware zero-shot object navigation (ZSON) approach designed for complex multi-room indoor environments. By planning hierarchically leveraging a global topologigal map with layout information and local imperative approach with detailed scene representation memory, ELA-ZSON achieves both efficient and effective navigation. The process is managed by an LLM-powered agent, ensuring seamless effective planning and navigation, without the need for human interaction, complex rewards, or costly training. Our experimental results on the MP3D benchmark achieves 85\% object navigation success rate (SR) and 79\% success rate weighted by path length (SPL) (over 40\% point improvement in SR and 60\% improvement in SPL compared to exsisting methods). Furthermore, we validate the robustness of our approach through virtual agent and real-world robotic deployment, showcasing its capability in practical scenarios. See https://anonymous.4open.science/r/ELA-ZSON-C67E/ for details.
The strong planning and reasoning capabilities of Large Language Models (LLMs) have fostered the development of agent-based systems capable of leveraging external tools and interacting with increasingly complex environments. However, these powerful features also introduce a critical security risk: indirect prompt injection, a sophisticated attack vector that compromises the core of these agents, the LLM, by manipulating contextual information rather than direct user prompts. In this work, we propose a generic black-box fuzzing framework, AgentXploit, designed to automatically discover and exploit indirect prompt injection vulnerabilities across diverse LLM agents. Our approach starts by constructing a high-quality initial seed corpus, then employs a seed selection algorithm based on Monte Carlo Tree Search (MCTS) to iteratively refine inputs, thereby maximizing the likelihood of uncovering agent weaknesses. We evaluate AgentXploit on two public benchmarks, AgentDojo and VWA-adv, where it achieves 71% and 70% success rates against agents based on o3-mini and GPT-4o, respectively, nearly doubling the performance of baseline attacks. Moreover, AgentXploit exhibits strong transferability across unseen tasks and internal LLMs, as well as promising results against defenses. Beyond benchmark evaluations, we apply our attacks in real-world environments, successfully misleading agents to navigate to arbitrary URLs, including malicious sites.