Procedural activity assistants potentially support humans in a variety of settings, from our daily lives, e.g., cooking or assembling flat-pack furniture, to professional situations, e.g., manufacturing or biological experiments. Despite its potential use cases, the system development tailored for such an assistant is still underexplored. In this paper, we propose a novel framework, called TAMA, a Tool-Augmented Multimodal Agent, for procedural activity understanding. TAMA enables interleaved multimodal reasoning by making use of multimedia-returning tools in a training-free setting. Our experimental result on the multimodal procedural QA dataset, ProMQA-Assembly, shows that our approach can improve the performance of vision-language models, especially GPT-5 and MiMo-VL. Furthermore, our ablation studies provide empirical support for the effectiveness of two features that characterize our framework, multimedia-returning tools and agentic flexible tool selection. We believe our proposed framework and experimental results facilitate the thinking with images paradigm for video and multimodal tasks, let alone the development of procedural activity assistants.

Financial fraud detection in real-world scenarios presents significant challenges due to the subtlety and dispersion of evidence across complex, multi-year financial disclosures. In this work, we introduce a novel multi-agent reasoning framework AuditAgent, enhanced with auditing domain expertise, for fine-grained evidence chain localization in financial fraud cases. Leveraging an expert-annotated dataset constructed from enforcement documents and financial reports released by the China Securities Regulatory Commission, our approach integrates subject-level risk priors, a hybrid retrieval strategy, and specialized agent modules to efficiently identify and aggregate cross-report evidence. Extensive experiments demonstrate that our method substantially outperforms General-Purpose Agent paradigm in both recall and interpretability, establishing a new benchmark for automated, transparent financial forensics. Our results highlight the value of domain-specific reasoning and dataset construction for advancing robust financial fraud detection in practical, real-world regulatory applications.

Despite rapid progress in autonomous robotics, executing complex or long-horizon tasks remains a fundamental challenge. Most current approaches follow an open-loop paradigm with limited reasoning and no feedback, resulting in poor robustness to environmental changes and severe error accumulation. We present RoboPilot, a dual-thinking closed-loop framework for robotic manipulation that supports adaptive reasoning for complex tasks in real-world dynamic environments. RoboPilot leverages primitive actions for structured task planning and flexible action generation, while introducing feedback to enable replanning from dynamic changes and execution errors. Chain-of-Thought reasoning further enhances high-level task planning and guides low-level action generation. The system dynamically switches between fast and slow thinking to balance efficiency and accuracy. To systematically evaluate the robustness of RoboPilot in diverse robot manipulation scenarios, we introduce RoboPilot-Bench, a benchmark spanning 21 tasks across 10 categories, including infeasible-task recognition and failure recovery. Experiments show that RoboPilot outperforms state-of-the-art baselines by 25.9\% in task success rate, and the real-world deployment on an industrial robot further demonstrates its robustness in real-world settings.

Machine unlearning is an emerging technique that removes the influence of a subset of training data (forget set) from a model without full retraining, with applications including privacy protection, content moderation, and model correction. The key challenge lies in ensuring that the model completely forgets the knowledge of the forget set without compromising its overall utility. Existing unlearning methods for large language models (LLMs) often utilize auxiliary language models, retain datasets, or even commercial AI services for effective unlearning and maintaining the model utility. However, dependence on these external resources is often impractical and could potentially introduce additional privacy risks. In this work, we propose direct token optimization (DTO), a novel self-contained unlearning approach for LLMs that directly optimizes the token level objectives and eliminates the need for external resources. Given a sequence to unlearn, we identify two categories of tokens: target tokens, which capture critical knowledge for unlearning, and the remaining non-target tokens, which are crucial for maintaining the model utility. The former are used to optimize the unlearning objective, while the latter serve to preserve the model's performance. The experimental results show that the proposed DTO achieves up to 16.8 improvement in forget quality on several benchmark datasets than the latest baselines while maintaining a comparable level of model utility. Machine unlearning aims to remove the effect of a subset of training data (referred to as the forget set) from a trained model (Cao & Y ang, 2015). The concept was introduced in response to data protection regulations such as General Data Protection Regulation (GDPR) (Mantelero, 2013), which established the'right to be forgotten'. A successfully unlearned model should fully eliminate the influence of forget set (unlearning efficacy), and preserve overall performance (model utility). Additionally, the unlearning algorithm should be more efficient than retraining (efficiency).

Large language models (LLMs) have been extensively used for legal judgment prediction tasks based on case reports and crime history. However, with a surge in the availability of large vision language models (VLMs), legal judgment prediction systems can now be made to leverage the images of the criminals in addition to the textual case reports/crime history. Applications built in this way could lead to inadvertent consequences and be used with malicious intent. In this work, we run an audit to investigate the efficiency of standalone VLMs in the bail decision prediction task. We observe that the performance is poor across multiple intersectional groups and models \textit{wrongly deny bail to deserving individuals with very high confidence}. We design different intervention algorithms by first including legal precedents through a RAG pipeline and then fine-tuning the VLMs using innovative schemes. We demonstrate that these interventions substantially improve the performance of bail prediction. Our work paves the way for the design of smarter interventions on VLMs in the future, before they can be deployed for real-world legal judgment prediction.

Foundation models have reshaped AI by unifying fragmented architectures into scalable backbones with multimodal reasoning and contextual adaptation. In parallel, the long-standing notion of AI agents, defined by the sensing-decision-action loop, is entering a new paradigm: with FMs as their cognitive core, agents transcend rule-based behaviors to achieve autonomy, generalization, and self-reflection. This dual shift is reinforced by real-world demands such as autonomous driving, robotics, virtual assistants, and GUI agents, as well as ecosystem advances in embedded hardware, edge computing, mobile deployment platforms, and communication protocols that together enable large-scale deployment. Yet this convergence collides with reality: while applications demand long-term adaptability and real-time interaction, mobile and edge deployments remain constrained by memory, energy, bandwidth, and latency. This creates a fundamental tension between the growing complexity of FMs and the limited resources of deployment environments. This survey provides the first systematic characterization of adaptive, resource-efficient agentic AI systems. We summarize enabling techniques into elastic inference, test-time adaptation, dynamic multimodal integration, and agentic AI applications, and identify open challenges in balancing accuracy-latency-communication trade-offs and sustaining robustness under distribution shifts. We further highlight future opportunities in algorithm-system co-design, cognitive adaptation, and collaborative edge deployment. By mapping FM structures, cognition, and hardware resources, this work establishes a unified perspective toward scalable, adaptive, and resource-efficient agentic AI. We believe this survey can help readers to understand the connections between enabling technologies while promoting further discussions on the fusion of agentic intelligence and intelligent agents.

We introduce Geo-R1, a reasoning-centric post-training framework that unlocks geospatial reasoning in vision-language models by combining thinking scaffolding and elevating. In the scaffolding stage, Geo-R1 instills a "geospatial thinking paradigm" via supervised fine-tuning on synthetic chain-of-thought exemplars, enabling models to connect visual cues with geographic priors without costly human reasoning annotations. In the elevating stage, it uses GRPO-based reinforcement learning on a weakly-supervised cross-view pairing proxy. This design supplies a verifiable and scalable reward signal: teaching models to capture and reconcile features across modalities, and harnessing reasoning for accurate prediction. Geo-R1 extends geospatial modeling from domain pretraining / supervised finetuning to reasoning-first post-training, and achieves state-of-the-art performance across various geospatial reasoning benchmarks. Our model is available at https://huggingface.co/miniHui/Geo-R1. Figure 1: Geo-R1 significantly outperforms baseline Bai et al. (2025) across 13 verifiable geo-reasoning tasks on the GeoChain benchmark (Y er-ramilli et al., 2025) in the zero-shot setting. See Table 6 for detailed description of these tasks. Geospatial reasoning is fundamental to a wide range of scientific and societal applications, spanning disaster response, search and rescue, urban planning, environmental monitoring, and sociocultural study. Unlike common vision-language reasoning (Li et al., 2024) centering around object recognition, captioning and general question-answering, geospatial reasoning spans many modalities (e.g., aerial imagery, streetview photos, location metadata, place information, etc.), and varied tasks (e.g., geographical, environmental, sociocultural, etc.) as shown in Figure 1. This blend of multimodal evidence and knowledge-intensive tasking makes general reasoning both crucial for geospatial understanding, and also uniquely challenging. While effective in natural domains, SFT is poorly suited in geospatial settings. Geospatial raw data can be plentiful, but supervisions are sparse, usually limited to coordinate metadata without descriptive content.

Abstract--The evolution of the 5S methodology with the support of artificial intelligence techniques represents a significant opportunity to improve industrial organization audits in the automotive chain, making them more objective, efficient and aligned with Industry 4.0 standards. This work developed an automated 5S audit system based on large-scale language models (LLM), capable of assessing the five senses (Seiri, Seiton, Seiso, Seiketsu, Shitsuke) in a standardized way through intelligent image analysis. The system's reliability was validated using Cohen's concordance coefficient (κ = 0.75), showing strong alignment between the automated assessments and the corresponding human audits. The results indicate that the proposed solution contributes significantly to continuous improvement in automotive manufacturing environments, speeding up the audit process by 50% of the traditional time and maintaining the consistency of the assessments, with a 99.8% reduction in operating costs compared to traditional manual audits. The methodology presented establishes a new paradigm for integrating lean systems with emerging AI technologies, offering scalability for implementation in automotive plants of different sizes. The global automotive industry faces growing competitiveness challenges demanding maximized operational efficiency and production quality. The 5S methodology, recognized worldwide as the foundation for workplace organization and cleanliness, plays a strategic role in operational excellence.

Multimodal Large Language Models (MLLMs) have made significant strides in visual understanding tasks. However, their performance on high-resolution images remains suboptimal. While existing approaches often attribute this limitation to perceptual constraints and argue that MLLMs struggle to recognize small objects, leading them to use "zoom in" strategies for better detail, our analysis reveals a different cause: the main issue is not object size, but rather caused by complex background interference. We systematically analyze this "zoom in" operation through a series of decoupling experiments and propose the Hierarchical Decoupling Framework (HiDe), a training-free framework that uses Token-wise Attention Decoupling (TAD) to decouple the question tokens and identify the key information tokens, then leverages their attention weights to achieve precise alignment with the target visual regions. Subsequently, it employs Layout-Preserving Decoupling (LPD) to decouple these regions from the background and reconstructs a compact representation that preserves essential spatial layouts while eliminating background interference. HiDe sets a new SOTA on V*Bench, HRBench4K, and HRBench8K, boosting Qwen2.5-VL 7B and InternVL3 8B to SOTA (92.1% and 91.6% on V*Bench), even surpassing RL methods. After optimization, HiDe uses 75% less memory than the previous training-free approach. Code is provided in https://github.com/Tennine2077/HiDe.

Vision-Language Models (VLMs) often produce fluent Natural Language Explanations (NLEs) that sound convincing but may not reflect the causal factors driving predictions. This mismatch of plausibility and faithfulness poses technical and governance risks. We introduce Explanation-Driven Counterfactual Testing (EDCT), a fully automated verification procedure for a target VLM that treats the model's own explanation as a falsifiable hypothesis. Given an image-question pair, EDCT: (1) obtains the model's answer and NLE, (2) parses the NLE into testable visual concepts, (3) generates targeted counterfactual edits via generative inpainting, and (4) computes a Counterfactual Consistency Score (CCS) using LLM-assisted analysis of changes in both answers and explanations. Across 120 curated OK-VQA examples and multiple VLMs, EDCT uncovers substantial faithfulness gaps and provides regulator-aligned audit artifacts indicating when cited concepts fail causal tests.