1. Compute as Teacher: Turning Inference Compute Into Reference-Free Supervision
Authors: Dulhan Jayalath, Shashwat Goel, Thomas Foster, Parag Jain, Suchin Gururangan, Cheng Zhang, Anirudh Goyal, Alan Schelten •
Published: 2025-09-17 •
Source: arXiv
Where do learning signals come from when there is no ground truth in post-training? We propose turning exploration into supervision through Compute as Teacher (CaT), which converts the model's own exploration at inference-time into reference-free supervision by synthesizing a single reference from a group of parallel rollouts and then optimizing toward it. Concretely, the current policy produces a group of rollouts; a frozen anchor (the initial policy) reconciles omissions and contradictions to estimate a reference, turning extra inference-time compute into a teacher signal. We turn this into rewards in two regimes: (i) verifiable tasks use programmatic equivalence on final answers; (ii) non-verifiable tasks use self-proposed rubrics-binary, auditable criteria scored by an independent LLM judge, with reward given by the fraction satisfied. Unlike selection methods (best-of-N, majority, perplexity, or judge scores), synthesis may disagree with the majority and be correct even when all rollouts are wrong; performance scales with the number of rollouts. As a test-time procedure, CaT improves Gemma 3 4B, Qwen 3 4B, and Llama 3.1 8B (up to +27% on MATH-500; +12% on HealthBench). With reinforcement learning (CaT-RL), we obtain further gains (up to +33% and +30%), with the trained policy surpassing the initial teacher signal.
2. A Universal Banach--Bregman Framework for Stochastic Iterations: Unifying Stochastic Mirror Descent, Learning and LLM Training
Authors: Johnny R. Zhang, Xiaomei Mi, Gaoyuan Du, Qianyi Sun, Shiqi Wang, Jiaxuan Li, Wenhua Zhou •
Published: 2025-09-17 •
Source: arXiv
Stochastic optimization powers the scalability of modern artificial intelligence, spanning machine learning, deep learning, reinforcement learning, and large language model training. Yet, existing theory remains largely confined to Hilbert spaces, relying on inner-product frameworks and orthogonality. This paradigm fails to capture non-Euclidean settings, such as mirror descent on simplices, Bregman proximal methods for sparse learning, natural gradient descent in information geometry, or Kullback--Leibler-regularized language model training. Unlike Euclidean-based Hilbert-space methods, this approach embraces general Banach spaces. This work introduces a pioneering Banach--Bregman framework for stochastic iterations, establishing Bregman geometry as a foundation for next-generation optimization. It (i) provides a unified template via Bregman projections and Bregman--Fejer monotonicity, encompassing stochastic approximation, mirror descent, natural gradient, adaptive methods, and mirror-prox; (ii) establishes super-relaxations ($\lambda > 2$) in non-Hilbert settings, enabling flexible geometries and elucidating their acceleration effect; and (iii) delivers convergence theorems spanning almost-sure boundedness to geometric rates, validated on synthetic and real-world tasks. Empirical studies across machine learning (UCI benchmarks), deep learning (e.g., Transformer training), reinforcement learning (actor--critic), and large language models (WikiText-2 with distilGPT-2) show up to 20% faster convergence, reduced variance, and enhanced accuracy over classical baselines. These results position Banach--Bregman geometry as a cornerstone unifying optimization theory and practice across core AI paradigms.
3. Framing Migration: A Computational Analysis of UK Parliamentary Discourse
Authors: Vahid Ghafouri, Robert McNeil, Teodor Yankov, Madeleine Sumption, Luc Rocher, Scott A. Hale, Adam Mahdi •
Published: 2025-09-17 •
Source: arXiv
We present a large-scale computational analysis of migration-related discourse in UK parliamentary debates spanning over 75 years and compare it with US congressional discourse. Using open-weight LLMs, we annotate each statement with high-level stances toward migrants and track the net tone toward migrants across time and political parties. For the UK, we extend this with a semi-automated framework for extracting fine-grained narrative frames to capture nuances of migration discourse. Our findings show that, while US discourse has grown increasingly polarised, UK parliamentary attitudes remain relatively aligned across parties, with a persistent ideological gap between Labour and the Conservatives, reaching its most negative level in 2025. The analysis of narrative frames in the UK parliamentary statements reveals a shift toward securitised narratives such as border control and illegal immigration, while longer-term integration-oriented frames such as social integration have declined. Moreover, discussions of national law about immigration have been replaced over time by international law and human rights, revealing nuances in discourse trends. Taken together broadly, our findings demonstrate how LLMs can support scalable, fine-grained discourse analysis in political and historical contexts.
4. MCGS-SLAM: A Multi-Camera SLAM Framework Using Gaussian Splatting for High-Fidelity Mapping
Authors: Zhihao Cao, Hanyu Wu, Li Wa Tang, Zizhou Luo, Zihan Zhu, Wei Zhang, Marc Pollefeys, Martin R. Oswald •
Published: 2025-09-17 •
Source: arXiv
Recent progress in dense SLAM has primarily targeted monocular setups, often at the expense of robustness and geometric coverage. We present MCGS-SLAM, the first purely RGB-based multi-camera SLAM system built on 3D Gaussian Splatting (3DGS). Unlike prior methods relying on sparse maps or inertial data, MCGS-SLAM fuses dense RGB inputs from multiple viewpoints into a unified, continuously optimized Gaussian map. A multi-camera bundle adjustment (MCBA) jointly refines poses and depths via dense photometric and geometric residuals, while a scale consistency module enforces metric alignment across views using low-rank priors. The system supports RGB input and maintains real-time performance at large scale. Experiments on synthetic and real-world datasets show that MCGS-SLAM consistently yields accurate trajectories and photorealistic reconstructions, usually outperforming monocular baselines. Notably, the wide field of view from multi-camera input enables reconstruction of side-view regions that monocular setups miss, critical for safe autonomous operation. These results highlight the promise of multi-camera Gaussian Splatting SLAM for high-fidelity mapping in robotics and autonomous driving.
5. Bridging Past and Future: Distribution-Aware Alignment for Time Series Forecasting
Authors: Yifan Hu, Jie Yang, Tian Zhou, Peiyuan Liu, Yujin Tang, Rong Jin, Liang Sun •
Published: 2025-09-17 •
Source: arXiv
Representation learning techniques like contrastive learning have long been explored in time series forecasting, mirroring their success in computer vision and natural language processing. Yet recent state-of-the-art (SOTA) forecasters seldom adopt these representation approaches because they have shown little performance advantage. We challenge this view and demonstrate that explicit representation alignment can supply critical information that bridges the distributional gap between input histories and future targets. To this end, we introduce TimeAlign, a lightweight, plug-and-play framework that learns auxiliary features via a simple reconstruction task and feeds them back to any base forecaster. Extensive experiments across eight benchmarks verify its superior performance. Further studies indicate that the gains arises primarily from correcting frequency mismatches between historical inputs and future outputs. We also provide a theoretical justification for the effectiveness of TimeAlign in increasing the mutual information between learned representations and predicted targets. As it is architecture-agnostic and incurs negligible overhead, TimeAlign can serve as a general alignment module for modern deep learning time-series forecasting systems. The code is available at https://github.com/TROUBADOUR000/TimeAlign.
6. Characterization of superconducting germanide and germanosilicide films of Pd, Pt, Rh and Ir formed by solid-phase epitaxy
Authors: Hao Li, Zhongxia Shang, Michael P. Lilly, Maksym Myronov, Leonid P. Rokhinson •
Published: 2025-09-17 •
Source: arXiv
Facilitated by recent advances in strained Ge/SiGe quantum well (QW) growth technology, superconductor-semiconductor hybrid devices based on group IV materials have been developed, potentially augmenting the functionality of quantum circuits. The formation of highly transparent superconducting platinum germanosilicide (PtSiGe) contacts to Ge/SiGe heterostructures by solid-phase epitaxy between Pt and SiGe has recently been reported, although with a relatively low critical temperature $<1\,\mathrm{K}$. Here, we present a comparative study of the superconducting properties of Pt, Pd, Rh, and Ir germanides, along with an in-depth characterization of Ir(Si)Ge films formed by solid-phase epitaxy. For films fabricated under optimal epitaxy conditions, we report $T_\mathrm{c}=3.4\,\mathrm{K}$ ($2.6\,\mathrm{K}$ for IrGe (IrSiGe). High-resolution scanning transmission electron microscopy (HRSTEM) and energy-dispersive X-ray spectroscopy (EDX) reveal that Ir reacts with Ge substrates to form a polycrystalline IrGe layer with a sharp IrGe/Ge interface.
7. TGPO: Tree-Guided Preference Optimization for Robust Web Agent Reinforcement Learning
Authors: Ziyuan Chen, Zhenghui Zhao, Zhangye Han, Miancan Liu, Xianhang Ye, Yiqing Li, Hongbo Min, Jinkui Ren, Xiantao Zhang, Guitao Cao •
Published: 2025-09-17 •
Source: arXiv
With the rapid advancement of large language models and vision-language models, employing large models as Web Agents has become essential for automated web interaction. However, training Web Agents with reinforcement learning faces critical challenges including credit assignment misallocation, prohibitively high annotation costs, and reward sparsity. To address these issues, we propose Tree-Guided Preference Optimization (TGPO), an offline reinforcement learning framework that proposes a tree-structured trajectory representation merging semantically identical states across trajectories to eliminate label conflicts. Our framework incorporates a Process Reward Model that automatically generates fine-grained rewards through subgoal progress, redundancy detection, and action verification. Additionally, a dynamic weighting mechanism prioritizes high-impact decision points during training. Experiments on Online-Mind2Web and our self-constructed C-WebShop datasets demonstrate that TGPO significantly outperforms existing methods, achieving higher success rates with fewer redundant steps.
8. AssoCiAm: A Benchmark for Evaluating Association Thinking while Circumventing Ambiguity
Authors: Yifan Liu, Wenkuan Zhao, Shanshan Zhong, Jinghui Qin, Mingfu Liang, Zhongzhan Huang, Wushao Wen •
Published: 2025-09-17 •
Source: arXiv
Recent advancements in multimodal large language models (MLLMs) have garnered significant attention, offering a promising pathway toward artificial general intelligence (AGI). Among the essential capabilities required for AGI, creativity has emerged as a critical trait for MLLMs, with association serving as its foundation. Association reflects a model' s ability to think creatively, making it vital to evaluate and understand. While several frameworks have been proposed to assess associative ability, they often overlook the inherent ambiguity in association tasks, which arises from the divergent nature of associations and undermines the reliability of evaluations. To address this issue, we decompose ambiguity into two types-internal ambiguity and external ambiguity-and introduce AssoCiAm, a benchmark designed to evaluate associative ability while circumventing the ambiguity through a hybrid computational method. We then conduct extensive experiments on MLLMs, revealing a strong positive correlation between cognition and association. Additionally, we observe that the presence of ambiguity in the evaluation process causes MLLMs' behavior to become more random-like. Finally, we validate the effectiveness of our method in ensuring more accurate and reliable evaluations. See Project Page for the data and codes.
9. TopoSizing: An LLM-aided Framework of Topology-based Understanding and Sizing for AMS Circuits
Authors: Ziming Wei, Zichen Kong, Yuan Wang, David Z. Pan, Xiyuan Tang •
Published: 2025-09-17 •
Source: arXiv
Analog and mixed-signal circuit design remains challenging due to the shortage of high-quality data and the difficulty of embedding domain knowledge into automated flows. Traditional black-box optimization achieves sampling efficiency but lacks circuit understanding, which often causes evaluations to be wasted in low-value regions of the design space. In contrast, learning-based methods embed structural knowledge but are case-specific and costly to retrain. Recent attempts with large language models show potential, yet they often rely on manual intervention, limiting generality and transparency. We propose TopoSizing, an end-to-end framework that performs robust circuit understanding directly from raw netlists and translates this knowledge into optimization gains. Our approach first applies graph algorithms to organize circuits into a hierarchical device-module-stage representation. LLM agents then execute an iterative hypothesis-verification-refinement loop with built-in consistency checks, producing explicit annotations. Verified insights are integrated into Bayesian optimization through LLM-guided initial sampling and stagnation-triggered trust-region updates, improving efficiency while preserving feasibility.
10. Deconstructing Intraocular Pressure: A Non-invasive Multi-Stage Probabilistic Inverse Framework
Authors: Md Rezwan Jaher, Abul Mukid Mohammad Mukaddes, A. B. M. Abdul Malek •
Published: 2025-09-17 •
Source: arXiv
Many critical healthcare decisions are challenged by the inability to measure key underlying parameters. Glaucoma, a leading cause of irreversible blindness driven by elevated intraocular pressure (IOP), provides a stark example. The primary determinant of IOP, a tissue property called trabecular meshwork permeability, cannot be measured in vivo, forcing clinicians to depend on indirect surrogates. This clinical challenge is compounded by a broader computational one: developing predictive models for such ill-posed inverse problems is hindered by a lack of ground-truth data and prohibitive cost of large-scale, high-fidelity simulations. We address both challenges with an end-to-end framework to noninvasively estimate unmeasurable variables from sparse, routine data. Our approach combines a multi-stage artificial intelligence architecture to functionally separate the problem; a novel data generation strategy we term PCDS that obviates the need for hundreds of thousands of costly simulations, reducing the effective computational time from years to hours; and a Bayesian engine to quantify predictive uncertainty. Our framework deconstructs a single IOP measurement into its fundamental components from routine inputs only, yielding estimates for the unmeasurable tissue permeability and a patient's outflow facility. Our noninvasively estimated outflow facility achieved excellent agreement with state-of-the-art tonography with precision comparable to direct physical instruments. Furthermore, the newly derived permeability biomarker demonstrates high accuracy in stratifying clinical cohorts by disease risk, highlighting its diagnostic potential. More broadly, our framework establishes a generalizable blueprint for solving similar inverse problems in other data-scarce, computationally-intensive domains.
11. MIMIC-D: Multi-modal Imitation for MultI-agent Coordination with Decentralized Diffusion Policies
Authors: Dayi Dong, Maulik Bhatt, Seoyeon Choi, Negar Mehr •
Published: 2025-09-17 •
Source: arXiv
As robots become more integrated in society, their ability to coordinate with other robots and humans on multi-modal tasks (those with multiple valid solutions) is crucial. We propose to learn such behaviors from expert demonstrations via imitation learning (IL). However, when expert demonstrations are multi-modal, standard IL approaches can struggle to capture the diverse strategies, hindering effective coordination. Diffusion models are known to be effective at handling complex multi-modal trajectory distributions in single-agent systems. Diffusion models have also excelled in multi-agent scenarios where multi-modality is more common and crucial to learning coordinated behaviors. Typically, diffusion-based approaches require a centralized planner or explicit communication among agents, but this assumption can fail in real-world scenarios where robots must operate independently or with agents like humans that they cannot directly communicate with. Therefore, we propose MIMIC-D, a Centralized Training, Decentralized Execution (CTDE) paradigm for multi-modal multi-agent imitation learning using diffusion policies. Agents are trained jointly with full information, but execute policies using only local information to achieve implicit coordination. We demonstrate in both simulation and hardware experiments that our method recovers multi-modal coordination behavior among agents in a variety of tasks and environments, while improving upon state-of-the-art baselines.
12. BEVUDA++: Geometric-aware Unsupervised Domain Adaptation for Multi-View 3D Object Detection
Authors: Rongyu Zhang, Jiaming Liu, Xiaoqi Li, Xiaowei Chi, Dan Wang, Li Du, Yuan Du, Shanghang Zhang •
Published: 2025-09-17 •
Source: arXiv
Vision-centric Bird's Eye View (BEV) perception holds considerable promise for autonomous driving. Recent studies have prioritized efficiency or accuracy enhancements, yet the issue of domain shift has been overlooked, leading to substantial performance degradation upon transfer. We identify major domain gaps in real-world cross-domain scenarios and initiate the first effort to address the Domain Adaptation (DA) challenge in multi-view 3D object detection for BEV perception. Given the complexity of BEV perception approaches with their multiple components, domain shift accumulation across multi-geometric spaces (e.g., 2D, 3D Voxel, BEV) poses a significant challenge for BEV domain adaptation. In this paper, we introduce an innovative geometric-aware teacher-student framework, BEVUDA++, to diminish this issue, comprising a Reliable Depth Teacher (RDT) and a Geometric Consistent Student (GCS) model. Specifically, RDT effectively blends target LiDAR with dependable depth predictions to generate depth-aware information based on uncertainty estimation, enhancing the extraction of Voxel and BEV features that are essential for understanding the target domain. To collaboratively reduce the domain shift, GCS maps features from multiple spaces into a unified geometric embedding space, thereby narrowing the gap in data distribution between the two domains. Additionally, we introduce a novel Uncertainty-guided Exponential Moving Average (UEMA) to further reduce error accumulation due to domain shifts informed by previously obtained uncertainty guidance. To demonstrate the superiority of our proposed method, we execute comprehensive experiments in four cross-domain scenarios, securing state-of-the-art performance in BEV 3D object detection tasks, e.g., 12.9\% NDS and 9.5\% mAP enhancement on Day-Night adaptation.
13. An Exploratory Study on Abstract Images and Visual Representations Learned from Them
Authors: Haotian Li, Jianbo Jiao •
Published: 2025-09-17 •
Source: arXiv
Imagine living in a world composed solely of primitive shapes, could you still recognise familiar objects? Recent studies have shown that abstract images-constructed by primitive shapes-can indeed convey visual semantic information to deep learning models. However, representations obtained from such images often fall short compared to those derived from traditional raster images. In this paper, we study the reasons behind this performance gap and investigate how much high-level semantic content can be captured at different abstraction levels. To this end, we introduce the Hierarchical Abstraction Image Dataset (HAID), a novel data collection that comprises abstract images generated from normal raster images at multiple levels of abstraction. We then train and evaluate conventional vision systems on HAID across various tasks including classification, segmentation, and object detection, providing a comprehensive study between rasterised and abstract image representations. We also discuss if the abstract image can be considered as a potentially effective format for conveying visual semantic information and contributing to vision tasks.
14. StableTracker: Learning to Stably Track Target via Differentiable Simulation
Authors: Fanxing Li, Shengyang Wang, Fangyu Sun, Shuyu Wu, Dexin Zuo, Wenxian Yu, Danping Zou •
Published: 2025-09-17 •
Source: arXiv
FPV object tracking methods heavily rely on handcraft modular designs, resulting in hardware overload and cumulative error, which seriously degrades the tracking performance, especially for rapidly accelerating or decelerating targets. To address these challenges, we present \textbf{StableTracker}, a learning-based control policy that enables quadrotors to robustly follow the moving target from arbitrary perspectives. The policy is trained using backpropagation-through-time via differentiable simulation, allowing the quadrotor to maintain the target at the center of the visual field in both horizontal and vertical directions, while keeping a fixed relative distance, thereby functioning as an autonomous aerial camera. We compare StableTracker against both state-of-the-art traditional algorithms and learning baselines. Simulation experiments demonstrate that our policy achieves superior accuracy, stability and generalization across varying safe distances, trajectories, and target velocities. Furthermore, a real-world experiment on a quadrotor with an onboard computer validated practicality of the proposed approach.
15. CLAW: A Vision-Language-Action Framework for Weight-Aware Robotic Grasping
Authors: Zijian An, Ran Yang, Yiming Feng, Lifeng Zhou •
Published: 2025-09-17 •
Source: arXiv
Vision-language-action (VLA) models have recently emerged as a promising paradigm for robotic control, enabling end-to-end policies that ground natural language instructions into visuomotor actions. However, current VLAs often struggle to satisfy precise task constraints, such as stopping based on numeric thresholds, since their observation-to-action mappings are implicitly shaped by training data and lack explicit mechanisms for condition monitoring. In this work, we propose CLAW (CLIP-Language-Action for Weight), a framework that decouples condition evaluation from action generation. CLAW leverages a fine-tuned CLIP model as a lightweight prompt generator, which continuously monitors the digital readout of a scale and produces discrete directives based on task-specific weight thresholds. These prompts are then consumed by $\pi_0$, a flow-based VLA policy, which integrates the prompts with multi-view camera observations to produce continuous robot actions. This design enables CLAW to combine symbolic weight reasoning with high-frequency visuomotor control. We validate CLAW on three experimental setups: single-object grasping and mixed-object tasks requiring dual-arm manipulation. Across all conditions, CLAW reliably executes weight-aware behaviors and outperforms both raw-$\pi_0$ and fine-tuned $\pi_0$ models. We have uploaded the videos as supplementary materials.
16. A proposal for automated turbulence modelling
Authors: Marco Castelletti, Maurizio Quadrio •
Published: 2025-09-17 •
Source: arXiv
Solving the Reynolds-averaged Navier-Stokes equations (RANS) closed with an eddy viscosity computed through a turbulence model is still the leading approach for Computational Fluid Dynamics simulations. Unfortunately, universal models with good predictive capabilities over a wide range of flows are not available. In this work, we propose the use of machine learning to improve existing RANS models. The approach does not require high-fidelity training data. A convolutional neural network is used to identify and segment at runtime the flow field into different zones, each resembling one item of a predefined list of elementary flows. The turbulence model applied in each zone is taken from an equally predefined set of classic models, each specifically tuned to work best for one elementary flow, free from the requirement of universality. The idea is first presented in general terms, and then demonstrated via a preliminary implementation, where only three elementary flows are considered, and three turbulence models are used. Test cases show that, already in this oversimplified form, automated zonal modelling outperforms the baseline RANS models without computational overhead.
17. Cybersecurity AI: Humanoid Robots as Attack Vectors
Authors: Víctor Mayoral-Vilches •
Published: 2025-09-17 •
Source: arXiv
We present a systematic security assessment of the Unitree G1 humanoid showing it operates simultaneously as a covert surveillance node and can be purposed as an active cyber operations platform. Partial reverse engineering of Unitree's proprietary FMX encryption reveal a static Blowfish-ECB layer and a predictable LCG mask-enabled inspection of the system's otherwise sophisticated security architecture, the most mature we have observed in commercial robotics. Two empirical case studies expose the critical risk of this humanoid robot: (a) the robot functions as a trojan horse, continuously exfiltrating multi-modal sensor and service-state telemetry to 43.175.228.18:17883 and 43.175.229.18:17883 every 300 seconds without operator notice, creating violations of GDPR Articles 6 and 13; (b) a resident Cybersecurity AI (CAI) agent can pivot from reconnaissance to offensive preparation against any target, such as the manufacturer's cloud control plane, demonstrating escalation from passive monitoring to active counter-operations. These findings argue for adaptive CAI-powered defenses as humanoids move into critical infrastructure, contributing the empirical evidence needed to shape future security standards for physical-cyber convergence systems.
18. GeoAware-VLA: Implicit Geometry Aware Vision-Language-Action Model
Authors: Ali Abouzeid, Malak Mansour, Zezhou Sun, Dezhen Song •
Published: 2025-09-17 •
Source: arXiv
Vision-Language-Action (VLA) models often fail to generalize to novel camera viewpoints, a limitation stemming from their difficulty in inferring robust 3D geometry from 2D images. We introduce GeoAware-VLA, a simple yet effective approach that enhances viewpoint invariance by integrating strong geometric priors into the vision backbone. Instead of training a visual encoder or relying on explicit 3D data, we leverage a frozen, pretrained geometric vision model as a feature extractor. A trainable projection layer then adapts these geometrically-rich features for the policy decoder, relieving it of the burden of learning 3D consistency from scratch. Through extensive evaluations on LIBERO benchmark subsets, we show GeoAware-VLA achieves substantial improvements in zero-shot generalization to novel camera poses, boosting success rates by over 2x in simulation. Crucially, these benefits translate to the physical world; our model shows a significant performance gain on a real robot, especially when evaluated from unseen camera angles. Our approach proves effective across both continuous and discrete action spaces, highlighting that robust geometric grounding is a key component for creating more generalizable robotic agents.
19. The Cybersecurity of a Humanoid Robot
Authors: Víctor Mayoral-Vilches •
Published: 2025-09-17 •
Source: arXiv
The rapid advancement of humanoid robotics presents unprecedented cybersecurity challenges that existing theoretical frameworks fail to adequately address. This report presents a comprehensive security assessment of a production humanoid robot platform, bridging the gap between abstract security models and operational vulnerabilities. Through systematic static analysis, runtime observation, and cryptographic examination, we uncovered a complex security landscape characterized by both sophisticated defensive mechanisms and critical vulnerabilities. Our findings reveal a dual-layer proprietary encryption system (designated FMX') that, while innovative in design, suffers from fundamental implementation flaws including the use of static cryptographic keys that enable offline configuration decryption. More significantly, we documented persistent telemetry connections transmitting detailed robot state information--including audio, visual, spatial, and actuator data--to external servers without explicit user consent or notification mechanisms. We operationalized a Cybersecurity AI agent on the Unitree G1 to map and prepare exploitation of its manufacturer's cloud infrastructure, illustrating how a compromised humanoid can escalate from covert data collection to active counter-offensive operations. We argue that securing humanoid robots requires a paradigm shift toward Cybersecurity AI (CAI) frameworks that can adapt to the unique challenges of physical-cyber convergence. This work contributes empirical evidence for developing robust security standards as humanoid robots transition from research curiosities to operational systems in critical domains.
20. A neuromorphic continuous soil monitoring system for precision irrigation
Authors: Mirco Tincani, Khaled Kerouch, Umberto Garlando, Mattia Barezzi, Alessandro Sanginario, Giacomo Indiveri, Chiara De Luca •
Published: 2025-09-17 •
Source: arXiv
Sensory processing at the edge requires ultra-low power stand-alone computing technologies. This is particularly true for modern agriculture and precision irrigation systems which aim to optimize water usage by monitoring key environmental observables continuously using distributed efficient embedded processing elements. Neuromorphic processing systems are emerging as a promising technology for extreme edge-computing applications that need to run on resource-constrained hardware. As such, they are a very good candidate for implementing efficient water management systems based on data measured from soil and plants, across large fields. In this work, we present a fully energy-efficient neuromorphic irrigation control system that operates autonomously without any need for data transmission or remote processing. Leveraging the properties of a biologically realistic spiking neural network, our system performs computation, and decision-making locally. We validate this approach using real-world soil moisture data from apple and kiwi orchards applied to a mixed-signal neuromorphic processor, and show that the generated irrigation commands closely match those derived from conventional methods across different soil depths. Our results show that local neuromorphic inference can maintain decision accuracy, paving the way for autonomous, sustainable irrigation solutions at scale.
21. Queen Detection in Beehives via Environmental Sensor Fusion for Low-Power Edge Computing
Authors: Chiara De Luca, Elisa Donati •
Published: 2025-09-17 •
Source: arXiv
Queen bee presence is essential for the health and stability of honeybee colonies, yet current monitoring methods rely on manual inspections that are labor-intensive, disruptive, and impractical for large-scale beekeeping. While recent audio-based approaches have shown promise, they often require high power consumption, complex preprocessing, and are susceptible to ambient noise. To overcome these limitations, we propose a lightweight, multimodal system for queen detection based on environmental sensor fusion-specifically, temperature, humidity, and pressure differentials between the inside and outside of the hive. Our approach employs quantized decision tree inference on a commercial STM32 microcontroller, enabling real-time, low-power edge computing without compromising accuracy. We show that our system achieves over 99% queen detection accuracy using only environmental inputs, with audio features offering no significant performance gain. This work presents a scalable and sustainable solution for non-invasive hive monitoring, paving the way for autonomous, precision beekeeping using off-the-shelf, energy-efficient hardware.
22. Machines are more productive than humans until they aren't, and vice versa
Authors: Riccardo Zanardelli •
Published: 2025-09-17 •
Source: arXiv
With the growth of artificial skills, organizations may increasingly confront with the problem of optimizing skill policy decisions guided by economic principles. This paper addresses the underlying complexity of this challenge by developing an in-silico framework based on Monte Carlo simulations grounded in empirical realism to analyze the economic impact of human and machine skills, individually or jointly deployed, in the execution of tasks presenting varying levels of complexity. Our results provide quantitative support for the established notions that automation tends to be the most economically-effective strategy for tasks characterized by low-to-medium generalization difficulty, while automation struggles to match the economic utility of human skills in more complex scenarios. Critically, our simulations highlight that combining human and machine skills can be the most effective strategy when a high level of generalization is required, but only if genuine augmentation is achieved. In contrast, when failing to realize this synergy, the human-machine policy is severely penalized by the inherent costs of its dual skill structure, causing it to destroy value and becoming the worst choice from an economic perspective. The takeaway for decision-makers is unambiguous: simply allocating human and machine skills to a task is insufficient, and a human-machine skill policy is neither a silver-bullet solution nor a low-risk compromise. Rather, it is a critical opportunity to boost competitiveness that demands a strong organizational commitment to enabling augmentation. Also, our findings show that improving the cost-effectiveness of machine skills over time, while useful, does not replace the fundamental need to focus on achieving augmentation.
23. Whole-body Motion Control of an Omnidirectional Wheel-Legged Mobile Manipulator via Contact-Aware Dynamic Optimization
Authors: Zong Chen, Shaoyang Li, Ben Liu, Min Li, Zhouping Yin, Yiqun Li •
Published: 2025-09-17 •
Source: arXiv
Wheel-legged robots with integrated manipulators hold great promise for mobile manipulation in logistics, industrial automation, and human-robot collaboration. However, unified control of such systems remains challenging due to the redundancy in degrees of freedom, complex wheel-ground contact dynamics, and the need for seamless coordination between locomotion and manipulation. In this work, we present the design and whole-body motion control of an omnidirectional wheel-legged quadrupedal robot equipped with a dexterous manipulator. The proposed platform incorporates independently actuated steering modules and hub-driven wheels, enabling agile omnidirectional locomotion with high maneuverability in structured environments. To address the challenges of contact-rich interaction, we develop a contact-aware whole-body dynamic optimization framework that integrates point-contact modeling for manipulation with line-contact modeling for wheel-ground interactions. A warm-start strategy is introduced to accelerate online optimization, ensuring real-time feasibility for high-dimensional control. Furthermore, a unified kinematic model tailored for the robot's 4WIS-4WID actuation scheme eliminates the need for mode switching across different locomotion strategies, improving control consistency and robustness. Simulation and experimental results validate the effectiveness of the proposed framework, demonstrating agile terrain traversal, high-speed omnidirectional mobility, and precise manipulation under diverse scenarios, underscoring the system's potential for factory automation, urban logistics, and service robotics in semi-structured environments.
24. LLM Agents for Interactive Workflow Provenance: Reference Architecture and Evaluation Methodology
Authors: Renan Souza, Timothy Poteet, Brian Etz, Daniel Rosendo, Amal Gueroudji, Woong Shin, Prasanna Balaprakash, Rafael Ferreira da Silva •
Published: 2025-09-17 •
Source: arXiv
Modern scientific discovery increasingly relies on workflows that process data across the Edge, Cloud, and High Performance Computing (HPC) continuum. Comprehensive and in-depth analyses of these data are critical for hypothesis validation, anomaly detection, reproducibility, and impactful findings. Although workflow provenance techniques support such analyses, at large scale, the provenance data become complex and difficult to analyze. Existing systems depend on custom scripts, structured queries, or static dashboards, limiting data interaction. In this work, we introduce an evaluation methodology, reference architecture, and open-source implementation that leverages interactive Large Language Model (LLM) agents for runtime data analysis. Our approach uses a lightweight, metadata-driven design that translates natural language into structured provenance queries. Evaluations across LLaMA, GPT, Gemini, and Claude, covering diverse query classes and a real-world chemistry workflow, show that modular design, prompt tuning, and Retrieval-Augmented Generation (RAG) enable accurate and insightful LLM agent responses beyond recorded provenance.
25. An Empirical Study on Failures in Automated Issue Solving
Authors: Simiao Liu, Fang Liu, Liehao Li, Xin Tan, Yinghao Zhu, Xiaoli Lian, Li Zhang •
Published: 2025-09-17 •
Source: arXiv
Automated issue solving seeks to autonomously identify and repair defective code snippets across an entire codebase. SWE-Bench has emerged as the most widely adopted benchmark for evaluating progress in this area. While LLM-based agentic tools show great promise, they still fail on a substantial portion of tasks. Moreover, current evaluations primarily report aggregate issue-solving rates, which obscure the underlying causes of success and failure, making it challenging to diagnose model weaknesses or guide targeted improvements. To bridge this gap, we first analyze the performance and efficiency of three SOTA tools, spanning both pipeline-based and agentic architectures, in automated issue solving tasks of SWE-Bench-Verified under varying task characteristics. Furthermore, to move from high-level performance metrics to underlying cause analysis, we conducted a systematic manual analysis of 150 failed instances. From this analysis, we developed a comprehensive taxonomy of failure modes comprising 3 primary phases, 9 main categories, and 25 fine-grained subcategories. Then we systematically analyze the distribution of the identified failure modes, the results reveal distinct failure fingerprints between the two architectural paradigms, with the majority of agentic failures stemming from flawed reasoning and cognitive deadlocks. Motivated by these insights, we propose a collaborative Expert-Executor framework. It introduces a supervisory Expert agent tasked with providing strategic oversight and course-correction for a primary Executor agent. This architecture is designed to correct flawed reasoning and break the cognitive deadlocks that frequently lead to failure. Experiments show that our framework solves 22.2% of previously intractable issues for a leading single agent. These findings pave the way for building more robust agents through diagnostic evaluation and collaborative design.
26. MAP: End-to-End Autonomous Driving with Map-Assisted Planning
Authors: Huilin Yin, Yiming Kan, Daniel Watzenig •
Published: 2025-09-17 •
Source: arXiv
In recent years, end-to-end autonomous driving has attracted increasing attention for its ability to jointly model perception, prediction, and planning within a unified framework. However, most existing approaches underutilize the online mapping module, leaving its potential to enhance trajectory planning largely untapped. This paper proposes MAP (Map-Assisted Planning), a novel map-assisted end-to-end trajectory planning framework. MAP explicitly integrates segmentation-based map features and the current ego status through a Plan-enhancing Online Mapping module, an Ego-status-guided Planning module, and a Weight Adapter based on current ego status. Experiments conducted on the DAIR-V2X-seq-SPD dataset demonstrate that the proposed method achieves a 16.6% reduction in L2 displacement error, a 56.2% reduction in off-road rate, and a 44.5% improvement in overall score compared to the UniV2X baseline, even without post-processing. Furthermore, it achieves top ranking in Track 2 of the End-to-End Autonomous Driving through V2X Cooperation Challenge of MEIS Workshop @CVPR2025, outperforming the second-best model by 39.5% in terms of overall score. These results highlight the effectiveness of explicitly leveraging semantic map features in planning and suggest new directions for improving structure design in end-to-end autonomous driving systems. Our code is available at https://gitee.com/kymkym/map.git
27. Ensemble of Pre-Trained Models for Long-Tailed Trajectory Prediction
Authors: Divya Thuremella, Yi Yang, Simon Wanna, Lars Kunze, Daniele De Martini •
Published: 2025-09-17 •
Source: arXiv
This work explores the application of ensemble modeling to the multidimensional regression problem of trajectory prediction for vehicles in urban environments. As newer and bigger state-of-the-art prediction models for autonomous driving continue to emerge, an important open challenge is the problem of how to combine the strengths of these big models without the need for costly re-training. We show how, perhaps surprisingly, combining state-of-the-art deep learning models out-of-the-box (without retraining or fine-tuning) with a simple confidence-weighted average method can enhance the overall prediction. Indeed, while combining trajectory prediction models is not straightforward, this simple approach enhances performance by 10% over the best prediction model, especially in the long-tailed metrics. We show that this performance improvement holds on both the NuScenes and Argoverse datasets, and that these improvements are made across the dataset distribution. The code for our work is open source.
28. Personalized Detection of Stress via hdrEEG: Linking Neuro-markers to Cortisol, HRV, and Self-Report
Authors: N. B. Maimon, Ganit Baruchin, Itamar Grotto, Nathan Intrator, Talya Zeimer, Ofir Chibotero, Efrat Danino •
Published: 2025-09-17 •
Source: arXiv
Chronic stress is a major risk factor for cognitive decline, neurodegenerative disease, and systemic health burden, underscoring the need for reliable individual-level biomarkers of stress reactivity. While cortisol, heart rate variability (HRV), and self-report measures are widely used, they provide limited insight into neural mechanisms. Here, we tested whether two single-channel hdrEEG biomarkers, ST4 and T2, serve as personalized indices of stress regulation by linking neural activity to validated physiological and subjective measures. We conducted two studies. Study 1 included 101 healthy adults (22-82 years) who completed questionnaires on resilience, burnout, and perceived stress, provided salivary cortisol, and underwent hdrEEG during resting, detection, n-back, lexical, emotional music, and startle tasks. Study 2 included 82 adults (19-42 years) who completed the State-Trait Anxiety Inventory, were monitored for HRV, and performed auditory, stress-inducing (job interview, arithmetic), and emotional tasks during hdrEEG recording. Across studies, ST4 reflected physiological arousal and cognitive strain, correlating positively with cortisol, subjective stress, and pulse pressure, and negatively with resilience and HRV indices. T2 showed a complementary profile, linking emotional and autonomic processes. T2 activity correlated with cortisol, heart rate, HRV measures, resilience, and trait anxiety, especially during lexical and emotional tasks. Together, ST4 and T2 capture distinct facets of the stress response: physiological arousal versus emotional-regulatory sensitivity. These findings highlight portable hdrEEG as a promising tool for personalized stress assessment, bridging neural, physiological, and subjective domains with implications for clinical and occupational monitoring.
29. InterKey: Cross-modal Intersection Keypoints for Global Localization on OpenStreetMap
Authors: Nguyen Hoang Khoi Tran, Julie Stephany Berrio, Mao Shan, Stewart Worrall •
Published: 2025-09-17 •
Source: arXiv
Reliable global localization is critical for autonomous vehicles, especially in environments where GNSS is degraded or unavailable, such as urban canyons and tunnels. Although high-definition (HD) maps provide accurate priors, the cost of data collection, map construction, and maintenance limits scalability. OpenStreetMap (OSM) offers a free and globally available alternative, but its coarse abstraction poses challenges for matching with sensor data. We propose InterKey, a cross-modal framework that leverages road intersections as distinctive landmarks for global localization. Our method constructs compact binary descriptors by jointly encoding road and building imprints from point clouds and OSM. To bridge modality gaps, we introduce discrepancy mitigation, orientation determination, and area-equalized sampling strategies, enabling robust cross-modal matching. Experiments on the KITTI dataset demonstrate that InterKey achieves state-of-the-art accuracy, outperforming recent baselines by a large margin. The framework generalizes to sensors that can produce dense structural point clouds, offering a scalable and cost-effective solution for robust vehicle localization.
30. How Fly Neural Perception Mechanisms Enhance Visuomotor Control of Micro Robots
Authors: Renyuan Liu, Haoting Zhou, Chuankai Fang, Qinbing Fu •
Published: 2025-09-17 •
Source: arXiv
Anyone who has tried to swat a fly has likely been frustrated by its remarkable agility.This ability stems from its visual neural perception system, particularly the collision-selective neurons within its small brain.For autonomous robots operating in complex and unfamiliar environments, achieving similar agility is highly desirable but often constrained by the trade-off between computational cost and performance.In this context, insect-inspired intelligence offers a parsimonious route to low-power, computationally efficient frameworks.In this paper, we propose an attention-driven visuomotor control strategy inspired by a specific class of fly visual projection neurons-the lobula plate/lobula column type-2 (LPLC2)-and their associated escape behaviors.To our knowledge, this represents the first embodiment of an LPLC2 neural model in the embedded vision of a physical mobile robot, enabling collision perception and reactive evasion.The model was simplified and optimized at 70KB in memory to suit the computational constraints of a vision-based micro robot, the Colias, while preserving key neural perception mechanisms.We further incorporated multi-attention mechanisms to emulate the distributed nature of LPLC2 responses, allowing the robot to detect and react to approaching targets both rapidly and selectively.We systematically evaluated the proposed method against a state-of-the-art locust-inspired collision detection model.Results showed that the fly-inspired visuomotor model achieved comparable robustness, at success rate of 96.1% in collision detection while producing more adaptive and elegant evasive maneuvers.Beyond demonstrating an effective collision-avoidance strategy, this work highlights the potential of fly-inspired neural models for advancing research into collective behaviors in insect intelligence.
31. Who Taught the Lie? Responsibility Attribution for Poisoned Knowledge in Retrieval-Augmented Generation
Authors: Baolei Zhang, Haoran Xin, Yuxi Chen, Zhuqing Liu, Biao Yi, Tong Li, Lihai Nie, Zheli Liu, Minghong Fang •
Published: 2025-09-17 •
Source: arXiv
Retrieval-Augmented Generation (RAG) integrates external knowledge into large language models to improve response quality. However, recent work has shown that RAG systems are highly vulnerable to poisoning attacks, where malicious texts are inserted into the knowledge database to influence model outputs. While several defenses have been proposed, they are often circumvented by more adaptive or sophisticated attacks. This paper presents RAGOrigin, a black-box responsibility attribution framework designed to identify which texts in the knowledge database are responsible for misleading or incorrect generations. Our method constructs a focused attribution scope tailored to each misgeneration event and assigns a responsibility score to each candidate text by evaluating its retrieval ranking, semantic relevance, and influence on the generated response. The system then isolates poisoned texts using an unsupervised clustering method. We evaluate RAGOrigin across seven datasets and fifteen poisoning attacks, including newly developed adaptive poisoning strategies and multi-attacker scenarios. Our approach outperforms existing baselines in identifying poisoned content and remains robust under dynamic and noisy conditions. These results suggest that RAGOrigin provides a practical and effective solution for tracing the origins of corrupted knowledge in RAG systems.
32. Generative Image Coding with Diffusion Prior
Authors: Jianhui Chang •
Published: 2025-09-17 •
Source: arXiv
As generative technologies advance, visual content has evolved into a complex mix of natural and AI-generated images, driving the need for more efficient coding techniques that prioritize perceptual quality. Traditional codecs and learned methods struggle to maintain subjective quality at high compression ratios, while existing generative approaches face challenges in visual fidelity and generalization. To this end, we propose a novel generative coding framework leveraging diffusion priors to enhance compression performance at low bitrates. Our approach employs a pre-optimized encoder to generate generalized compressed-domain representations, integrated with the pretrained model's internal features via a lightweight adapter and an attentive fusion module. This framework effectively leverages existing pretrained diffusion models and enables efficient adaptation to different pretrained models for new requirements with minimal retraining costs. We also introduce a distribution renormalization method to further enhance reconstruction fidelity. Extensive experiments show that our method (1) outperforms existing methods in visual fidelity across low bitrates, (2) improves compression performance by up to 79% over H.266/VVC, and (3) offers an efficient solution for AI-generated content while being adaptable to broader content types.
33. Reinforcement Learning for Robotic Insertion of Flexible Cables in Industrial Settings
Authors: Jeongwoo Park, Seabin Lee, Changmin Park, Wonjong Lee, Changjoo Nam •
Published: 2025-09-17 •
Source: arXiv
The industrial insertion of flexible flat cables (FFCs) into receptacles presents a significant challenge owing to the need for submillimeter precision when handling the deformable cables. In manufacturing processes, FFC insertion with robotic manipulators often requires laborious human-guided trajectory generation. While Reinforcement Learning (RL) offers a solution to automate this task without modeling complex properties of FFCs, the nondeterminism caused by the deformability of FFCs requires significant efforts and time on training. Moreover, training directly in a real environment is dangerous as industrial robots move fast and possess no safety measure. We propose an RL algorithm for FFC insertion that leverages a foundation model-based real-to-sim approach to reduce the training time and eliminate the risk of physical damages to robots and surroundings. Training is done entirely in simulation, allowing for random exploration without the risk of physical damages. Sim-to-real transfer is achieved through semantic segmentation masks which leave only those visual features relevant to the insertion tasks such as the geometric and spatial information of the cables and receptacles. To enhance generality, we use a foundation model, Segment Anything Model 2 (SAM2). To eleminate human intervention, we employ a Vision-Language Model (VLM) to automate the initial prompting of SAM2 to find segmentation masks. In the experiments, our method exhibits zero-shot capabilities, which enable direct deployments to real environments without fine-tuning.
34. Inject, Fork, Compare: Defining an Interaction Vocabulary for Multi-Agent Simulation Platforms
Authors: HwiJoon Lee, Martina Di Paola, Yoo Jin Hong, Quang-Huy Nguyen, Joseph Seering •
Published: 2025-09-17 •
Source: arXiv
LLM-based multi-agent simulations are a rapidly growing field of research, but current simulations often lack clear modes for interaction and analysis, limiting the "what if" scenarios researchers are able to investigate. In this demo, we define three core operations for interacting with multi-agent simulations: inject, fork, and compare. Inject allows researchers to introduce external events at any point during simulation execution. Fork creates independent timeline branches from any timestamp, preserving complete state while allowing divergent exploration. Compare facilitates parallel observation of multiple branches, revealing how different interventions lead to distinct emergent behaviors. Together, these operations establish a vocabulary that transforms linear simulation workflows into interactive, explorable spaces. We demonstrate this vocabulary through a commodity market simulation with fourteen AI agents, where researchers can inject contrasting events and observe divergent outcomes across parallel timelines. By defining these fundamental operations, we provide a starting point for systematic causal investigation in LLM-based agent simulations, moving beyond passive observation toward active experimentation.
35. Mind the Gap: Aligning Knowledge Bases with User Needs to Enhance Mental Health Retrieval
Authors: Amanda Chan, James Jiayu Liu, He Kai, Onno P. Kampman •
Published: 2025-09-17 •
Source: arXiv
Access to reliable mental health information is vital for early help-seeking, yet expanding knowledge bases is resource-intensive and often misaligned with user needs. This results in poor performance of retrieval systems when presented concerns are not covered or expressed in informal or contextualized language. We present an AI-based gap-informed framework for corpus augmentation that authentically identifies underrepresented topics (gaps) by overlaying naturalistic user data such as forum posts in order to prioritize expansions based on coverage and usefulness. In a case study, we compare Directed (gap-informed augmentations) with Non-Directed augmentation (random additions), evaluating the relevance and usefulness of retrieved information across four retrieval-augmented generation (RAG) pipelines. Directed augmentation achieved near-optimal performance with modest expansions--requiring only a 42% increase for Query Transformation, 74% for Reranking and Hierarchical, and 318% for Baseline--to reach ~95% of the performance of an exhaustive reference corpus. In contrast, Non-Directed augmentation required substantially larger and thus practically infeasible expansions to achieve comparable performance (232%, 318%, 403%, and 763%, respectively). These results show that strategically targeted corpus growth can reduce content creation demands while sustaining high retrieval and provision quality, offering a scalable approach for building trusted health information repositories and supporting generative AI applications in high-stakes domains.