arXiv subjects

Qian Zhu

Publications and source records attributed to Qian Zhu.

At least 19 recordsLinked to original sources

arXiv2026

PrimeSim: A Generalist World Model Simulator from Large-Scale Cross-Embodiment Demonstrations

A robot world model that serves as a simulator must do more than generate plausible videos: it must faithfully follow commanded actions, maintain consistency across camera views, and sustain coherent long-horizon rollouts. We present PrimeSim, an action-conditioned, multiview world model simulator that addresses these challenges through progressive midtraining and posttraining. Starting from a pretrained video generator, we midtrain on 30,110 hours of robot demonstrations spanning 22 platforms. To improve action following, we introduce a unified cross-embodiment action representation and frame-aligned dual-path conditioning to strengthen action-frame correspondence. To maintain multiview consistency, we develop view-aware interleaved attention with explicit camera identities, enabling cross-view information exchange while preserving view-specific representations. We further introduce a fast-video, slow-memory posttraining framework that transforms the bidirectional video generator into an efficient autoregressive simulator. By combining fast causal generation, self-rollout training, and hierarchical temporal memory, PrimeSim mitigates error accumulation from self-generated histories and supports coherent long-horizon simulation. PrimeSim runs at 10.1 FPS, remains stable over long-horizon rollouts on rigid and deformable manipulation tasks, and transfers to held-out platforms within 100 to 200 gradient steps where the pretrained backbone does not. Against five open-source action-conditioned world models it leads four of the five WorldSimProbe suites and posts the best overall EWMBench score, while fine-grained local calibration still trails the field; its rollouts are faithful enough to support GRPO-based policy optimization. We release the model weights, inference code, and training code.

cs.RO↗

arXiv2026

Multi-AUV Ad-hoc Networks-Based Multi-Target Tracking Based on Scene-Adaptive Embodied Intelligence

With the rapid advancement of underwater net-working and multi-agent coordination technologies, autonomous underwater vehicle (AUV) ad-hoc networks have emerged as a pivotal framework for executing complex maritime missions, such as multi-target tracking. However, traditional data-centricarchitectures struggle to maintain operational consistency under highly dynamic topological fluctuations and severely constrained acoustic communication bandwidth. This article proposes a scene-adaptive embodied intelligence (EI) architecture for multi-AUV ad-hoc networks, which re-envisions AUVs as embodied entities by integrating perception, decision-making, and physical execution into a unified cognitive loop. To materialize the functional interaction between these layers, we define a beacon-based communication and control model that treats the communication link as a dynamic constraint-aware channel, effectively bridging the gap between high-level policy inference and decentralized physical actuation. Specifically, the proposed architecture employs a three-layer functional framework and introduces a Scene-Adaptive MARL (SA-MARL) algorithm featuring a dual-path critic mechanism. By integrating a scene critic network and a general critic network through a weight-based dynamic fusion process, SA-MARL effectively decouples specialized tracking tasks from global safety constraints, facilitating autonomous policy evolution. Evaluation results demonstrate that the proposedscheme significantly accelerates policy convergence and achieves superior tracking accuracy compared to mainstream MARL approaches, maintaining robust performance even under intense environmental interference and fluid topological shifts.

cs.RO↗

arXiv2026

Is That Clear? The Impacts of Clarifying Questions on Human-LLM Interaction

LLM assistants increasingly ask clarifying questions (CQs) to elicit preferences and tailor recommendations. Although prior work emphasizes improvements in answer quality and user experience, how CQs shape users' subsequent engagement and reliance remains unclear. We conducted a controlled experiment (N=180) across six everyday decision scenarios, comparing human-LLM interaction with and without CQs. We further manipulated whether responses violated explicit constraints and whether violations occurred in the sub-decisions the CQs addressed, while preserving elicited preferences. CQs increased perceived understanding and satisfaction, yet participants spent less time reviewing and evaluating responses and accepted more responses unchanged, including erroneous ones. Within the CQ condition, participants more often accepted responses containing CQ-adjacent violations than nonadjacent violations. These findings demonstrate that LLM's proactive CQs shape subsequent engagement and reliance with preference elicitation. We discuss implications for designing and evaluating proactive assistants that support continued user scrutiny of answers alongside personalization.

cs.HC↗

arXiv2026

ALIGN: Aligning Condition Estimation and Expression in Advantage-Conditioned Behavior Cloning

Human-collected robot manipulation data often mix successful behavior with retries, pauses, recoveries, and failures. Advantage-Conditioned Behavior Cloning (AC-BC) can exploit such mixed-quality data, but only if the conditioning signal is both accurately constructed from offline trajectories and effectively incorporated into the policy's action-vector-field prediction. Existing pipelines can weaken this signal by thresholding estimator-derived advantage scores into noisy conditions, and by routing conditions through prompts that weakly affect the action vector field. We introduce ALIGN, a condition-aligned AC-BC framework for vision-language-action policies. ALIGN directly predicts categorical pairwise conditions in the policy-facing label space and injects them into a DiT action expert through AdaRMS modulation fused with flow time. We evaluate ALIGN on deliberately mixed real-robot datasets for ARX X5 bowl stacking and Franka key hanging, each combining standard demonstrations with retry-heavy trajectories. ALIGN improves condition fidelity, strengthens conditional vector-field separability, and achieves 90.0% and 56.7% closed-loop success, compared with 56.7% and 33.3% for the strongest AC-BC baseline. These results indicate that aligning conditions from estimation to action prediction is an effective way to extract robust manipulation policies from data in which success, recovery and error coexist.

cs.RO↗

arXiv2026

REACT: Rolling Denoising and Dual Decoupling for Reactive Robot Control with VLA Models

Flow-based vision-language-action (VLA) models generate action chunks for temporally coherent robot motion, but chunked control creates a fundamental closed-loop trade-off: long chunks provide smooth execution, whereas frequent replanning improves reactivity at the cost of action discontinuities. We introduce REACT, a rolling-denoising framework that makes flow-based VLAs more reactive while preserving long-horizon context. Instead of regenerating entire action chunks from scratch, REACT maintains a persistent action buffer with staggered flow timesteps. At each control step, the full horizon is denoised using the latest observation, the cleanest action block is executed, partially refined future blocks are shifted forward, and fresh noise is appended to the tail. As a result, each executed action block is refined across multiple recent observations before deployment. To support real-time control, we further introduce dual decoupling, which separates sensing, VLM encoding, DiT denoising, and action execution, enabling high-frequency observation updates and action streaming under practical compute constraints. Across the RoboTwin 2.0 simulation benchmark and real-world tasks spanning bimanual manipulation and dynamic control on multiple robot platforms, REACT improves task success and reduces reaction latency while producing smoother trajectories than frequent-replanning and asynchronous baselines.

cs.RO↗

arXiv2026

Earth Surface Immune System for Rapid Monitoring of Unknown Anomalies

Earth surface anomalies, driven by escalating climate change, and expanding human activities, are increasing in both frequency and diversity, yet their limited historical data and unpredictability make them fundamentally different from conventional remote sensing targets. Existing methods address specific anomaly categories or stop at localization, leaving a gap between detection and actionable information. Here we present ESIA, an Earth Surface Immune System whose architecture is constrained by three principles from the biological immune system, refined over millions of years against equally diverse and uncertain threats. A non-specific innate immune stage treats anomalies as unobserved changes in time-series satellite imagery, generating binary localization maps at 14.51 km2/s without assuming any anomaly category, surpassing the strongest general baseline by 37% in F1. A specific adaptive immune stage applies negative selection to filter text prompts and matches surviving prompts with localized image patches through a multi-modal foundation model, enabling open-vocabulary recognition of unknown anomaly attributes including category, affected area, and damage severity, with recognition F1 exceeding 80%. A mutation mechanism tunes minimal embeddings at test time, adapting to each scene in 3.26s using a single reference image pair. We validate ESIA on a global-scale dataset covering 19,801.60 km2 across six anomaly categories, comparing against 22 models, and further apply it to quantify degraded farmland in the Dnipro Delta following the Kakhovka Dam collapse and assess burn severity from 2025 Palisades Fire in Los Angeles. This unprecedented flexibility in handling unknown anomalies opens new avenues for real-time disaster response and environmental surveillance.

cs.CV↗

arXiv2026

ForensicZoom: Adaptive Visual Inspection with Multimodal LLMs for Industrial-Grade Face Forgery Detection

Reliable face forgery detection is critical to the security of online identity verification systems, where missed attacks compromise security and excessive false positives disrupt legitimate users. Specialized forensic detectors achieve strong detection performance but provide limited interpretability, while multimodal large language models (MLLMs) offer strong semantic understanding and interpretable reasoning yet remain substantially weaker for face forgery detection. We argue that a key limitation lies in how visual evidence is acquired: subtle forensic artifacts may be poorly represented at standard resolution, while uniformly processing all cases at higher resolution is computationally inefficient. We therefore introduce ForensicZoom, an industrial-grade MLLM framework for adaptive visual inspection. ForensicZoom first equips a general-purpose MLLM with forensic-aware visual representations and aligns the language model with these features. Its central mechanism, NEED_ZOOM, enables the model to autonomously request magnified views of suspicious regions when the initial evidence is insufficient, turning fixed-pass classification into adaptive multi-round forensic reasoning. The zoom behavior is learned through reward shaping that balances detection accuracy with unnecessary visual inspection, concentrating additional computation on difficult cases. A final attribution optimization stage improves natural-language forensic reports while preserving detection performance. On large-scale industrial identity verification data, ForensicZoom achieves over 97% TPR at 0.1% FPR, substantially outperforming both specialized detectors and existing MLLM-based methods while producing actionable forensic attributions. These results demonstrate that ForensicZoom can provide an effective path toward accurate, interpretable, and scalable MLLM-based face forgery detection.

cs.CV↗

arXiv2026

Multi-AUV Ad-hoc network-based Target Tracking: A Value Gradient Guidance Multi-Agent Diffusion Reinforcement Learning Approach

Multi-AUV ad-hoc network-based target tracking requires networked autonomous underwater vehicles (AUVs) to cooperatively track maneuvering targets under constrained acoustic communication, dynamic topology, and uncertain ocean disturbances. Although multi-agent reinforcement learning (MARL) enables decentralized coordination through centralized training, existing methods suffer from high-dimensional joint state-action modeling, noise-sensitive policy generation, leading to unstable training and degraded tracking. To address these issues, we propose VGG-MADiffRL, a value-gradient-guided multi-agent diffusion RL algorithm, and MDCA, a diffusion?based hierarchical control architecture. Leveraging underwater mission characteristics, we model sonar detection mechanisms and ocean current disturbances, formulating cooperative tracking for multi-AUV ad-hoc networks as an MDP. The proposed MDCA constitutes a three-tier closed-loop control framework: a global intelligent control layer, a local online training layer, and a physical action execution layer. This structure enables synergistic optimization across task allocation, local decision processes, and execution feedback. Within MDCA, the local online training layer is the policy learning framework; VGG-MADiffRL builds on diffusion policies and incorporates value gradients to guide action generation in the reverse denoising process, steering the generated actions towards higher expected returns. It employs twin value networks with joint optimization and soft target updates to mitigate overestimation and training oscillations, promoting more stable convergence. Experimental results show that VGG-MADiffRL consistently achieves faster convergence, higher tracking accuracy, and smoother training dynamics in cooperative tracking scenarios, validating its effectiveness and practical engineering value in dynamic underwater settings.

cs.LG↗

arXiv2026

A Pragmatic VLA Foundation Model

Offering great potential in robotic manipulation, a capable Vision-Language-Action (VLA) foundation model is expected to faithfully generalize across tasks and platforms while ensuring cost efficiency (e.g., data and GPU hours required for adaptation). To this end, we develop LingBot-VLA with around 20,000 hours of real-world data from 9 popular dual-arm robot configurations. Through a systematic assessment on 4 robotic platforms, each completing 100 tasks with 130 post-training episodes per task, our model achieves clear superiority over competitors, showcasing its strong performance and broad generalizability. We have also built an efficient codebase, which delivers a throughput of 261 samples per second with an 8-GPU training setup, representing a 1.5~2.8$\times$ (depending on the relied VLM base model) speedup over existing VLA-oriented codebases. The above features ensure that our model is well-suited for real-world deployment. To advance the field of robot learning, we provide open access to the code, base model, and benchmark data, with a focus on enabling more challenging tasks and promoting sound evaluation standards.

cs.RO↗

arXiv2026

An Industrial-Scale Insurance LLM Achieving Verifiable Domain Mastery and Hallucination Control without Competence Trade-offs

Adapting Large Language Models (LLMs) to high-stakes vertical domains like insurance presents a significant challenge: scenarios demand strict adherence to complex regulations and business logic with zero tolerance for hallucinations. Existing approaches often suffer from a Competency Trade-off - sacrificing general intelligence for domain expertise - or rely heavily on RAG without intrinsic reasoning. To bridge this gap, we present INS-S1, an insurance-specific LLM family trained via a novel end-to-end alignment paradigm. Our approach features two methodological innovations: (1) A Verifiable Data Synthesis System that constructs hierarchical datasets for actuarial reasoning and compliance; and (2) A Progressive SFT-RL Curriculum Framework that integrates dynamic data annealing with a synergistic mix of Verified Reasoning (RLVR) and AI Feedback (RLAIF). By optimizing data ratios and reward signals, this framework enforces domain constraints while preventing catastrophic forgetting. Additionally, we release INSEva, the most comprehensive insurance benchmark to date (39k+ samples). Extensive experiments show that INS-S1 achieves SOTA performance on domain tasks, significantly outperforming DeepSeek-R1 and Gemini-2.5-Pro. Crucially, it maintains top-tier general capabilities and achieves a record-low 0.6% hallucination rate (HHEM). Our results demonstrate that rigorous domain specialization can be achieved without compromising general intelligence.

cs.CL↗

arXiv2026

DHEA-MECD: An Embodied Intelligence-Powered DRL Algorithm for AUV Tracking in Underwater Environments with High-Dimensional Features

In recent years, autonomous underwater vehicle (AUV) systems have demonstrated significant potential in complex marine exploration. However, effective AUV-based tracking remains challenging in realistic underwater environments characterized by high-dimensional features, including coupled kinematic states, spatial constraints, time-varying environmental disturbances, etc. To address these challenges, this paper proposes a hierarchical embodied-intelligence (EI) architecture for underwater multi-target tracking with AUVs in complex underwater environments. Built upon this architecture, we introduce the Double-Head Encoder-Attention-based Multi-Expert Collaborative Decision (DHEA-MECD), a novel Deep Reinforcement Learning (DRL) algorithm designed to support efficient and robust multi-target tracking. Specifically, in DHEA-MECD, a Double-Head Encoder-Attention-based information extraction framework is designed to semantically decompose raw sensory observations and explicitly model complex dependencies among heterogeneous features, including spatial configurations, kinematic states, structural constraints, and stochastic perturbations. On this basis, a motion-stage-aware multi-expert collaborative decision mechanism with Top-k expert selection strategy is introduced to support stage-adaptive decision-making. Furthermore, we propose the DHEA-MECD-based underwater multitarget tracking algorithm to enable AUV smart, stable, and anti-interference multi-target tracking. Extensive experimental results demonstrate that the proposed approach achieves superior tracking success rates, faster convergence, and improved motion optimality compared with mainstream DRL-based methods, particularly in complex and disturbance-rich marine environments.

cs.NI↗

arXiv2026

The Great March 100: 100 Detail-oriented Tasks for Evaluating Embodied AI Agents

Recently, with the rapid development of robot learning and imitation learning, numerous datasets and methods have emerged. However, these datasets and their task designs often lack systematic consideration and principles. This raises important questions: Do the current datasets and task designs truly advance the capabilities of robotic agents? Do evaluations on a few common tasks accurately reflect the differentiated performance of various methods proposed by different teams and evaluated on different tasks? To address these issues, we introduce the Great March 100 (\textbf{GM-100}) as the first step towards a robot learning Olympics. GM-100 consists of 100 carefully designed tasks that cover a wide range of interactions and long-tail behaviors, aiming to provide a diverse and challenging set of tasks to comprehensively evaluate the capabilities of robotic agents and promote diversity and complexity in robot dataset task designs. These tasks are developed through systematic analysis and expansion of existing task designs, combined with insights from human-object interaction primitives and object affordances. We collect a large amount of trajectory data on different robotic platforms and evaluate several baseline models. Experimental results demonstrate that the GM-100 tasks are 1) feasible to execute and 2) sufficiently challenging to effectively differentiate the performance of current VLA models. Our data and code are available at https://rhos.ai/research/gm-100.

cs.RO↗

arXiv2025

Ego vs. Exo and Active vs. Passive: Investigating the Individual and Combined Effects of Viewpoint and Navigation on Spatial Immersion and Understanding in Immersive Storytelling

Visual storytelling combines visuals and narratives to communicate important insights. While web-based visual storytelling is well-established, leveraging the next generation of digital technologies for visual storytelling, specifically immersive technologies, remains underexplored. We investigated the impact of the story viewpoint (from the audience's perspective) and navigation (when progressing through the story) on spatial immersion and understanding. First, we collected web-based 3D stories and elicited design considerations from three VR developers. We then adapted four selected web-based stories to an immersive format. Finally, we conducted a user study (N=24) to examine egocentric and exocentric viewpoints, active and passive navigation, and the combinations they form. Our results indicated significantly higher preferences for egocentric+active (higher agency and engagement) and exocentric+passive (higher focus on content). We also found a marginal significance of viewpoints on story understanding and a strong significance of navigation on spatial immersion.

cs.HC↗

arXiv2025

EPSO: A Caching-Based Efficient Superoptimizer for BPF Bytecode

Extended Berkeley Packet Filter (eBPF) allows developers to extend Linux kernel functionality without modifying its source code. To ensure system safety, an in-kernel safety checker, the verifier, enforces strict safety constraints (for example, a limited program size) on eBPF programs loaded into the kernel. These constraints, combined with eBPF's performance-critical use cases, make effective optimization essential. However, existing compilers (such as Clang) offer limited optimization support, and many semantics-preserving transformations are rejected by the verifier, which makes handcrafted optimization rule design both challenging and limited in effectiveness. Superoptimization overcomes the limitations of rule-based methods by automatically discovering optimal transformations, but its high computational cost limits scalability. To address this, we propose EPSO, a caching-based superoptimizer that discovers rewrite rules via offline superoptimization and reuses them to achieve high-quality optimizations with minimal runtime overhead. We evaluate EPSO on benchmarks from the Linux kernel and several eBPF-based projects, including Cilium, Katran, hXDP, Sysdig, Tetragon, and Tracee. EPSO discovers 795 rewrite rules and achieves up to 68.87 percent (average 24.37 percent) reduction in program size compared to Clang's output, outperforming the state-of-the-art BPF optimizer K2 on all benchmarks and Merlin on 92.68 percent of them. Additionally, EPSO reduces program runtime by an average of 6.60 percent, improving throughput and lowering latency in network applications.

cs.SE↗

arXiv2025

When Person Re-Identification Meets Event Camera: A Benchmark Dataset and An Attribute-guided Re-Identification Framework

Recent researchers have proposed using event cameras for person re-identification (ReID) due to their promising performance and better balance in terms of privacy protection, event camera-based person ReID has attracted significant attention. Currently, mainstream event-based person ReID algorithms primarily focus on fusing visible light and event stream, as well as preserving privacy. Although significant progress has been made, these methods are typically trained and evaluated on small-scale or simulated event camera datasets, making it difficult to assess their real identification performance and generalization ability. To address the issue of data scarcity, this paper introduces a large-scale RGB-event based person ReID dataset, called EvReID. The dataset contains 118,988 image pairs and covers 1200 pedestrian identities, with data collected across multiple seasons, scenes, and lighting conditions. We also evaluate 15 state-of-the-art person ReID algorithms, laying a solid foundation for future research in terms of both data and benchmarking. Based on our newly constructed dataset, this paper further proposes a pedestrian attribute-guided contrastive learning framework to enhance feature learning for person re-identification, termed TriPro-ReID. This framework not only effectively explores the visual features from both RGB frames and event streams, but also fully utilizes pedestrian attributes as mid-level semantic features. Extensive experiments on the EvReID dataset and MARS datasets fully validated the effectiveness of our proposed RGB-Event person ReID framework. The benchmark dataset and source code will be released on https://github.com/Event-AHU/Neuromorphic_ReID

cs.CV↗

arXiv2025

INSEva: A Comprehensive Chinese Benchmark for Large Language Models in Insurance

Insurance, as a critical component of the global financial system, demands high standards of accuracy and reliability in AI applications. While existing benchmarks evaluate AI capabilities across various domains, they often fail to capture the unique characteristics and requirements of the insurance domain. To address this gap, we present INSEva, a comprehensive Chinese benchmark specifically designed for evaluating AI systems' knowledge and capabilities in insurance. INSEva features a multi-dimensional evaluation taxonomy covering business areas, task formats, difficulty levels, and cognitive-knowledge dimension, comprising 38,704 high-quality evaluation examples sourced from authoritative materials. Our benchmark implements tailored evaluation methods for assessing both faithfulness and completeness in open-ended responses. Through extensive evaluation of 8 state-of-the-art Large Language Models (LLMs), we identify significant performance variations across different dimensions. While general LLMs demonstrate basic insurance domain competency with average scores above 80, substantial gaps remain in handling complex, real-world insurance scenarios. The benchmark will be public soon.

cs.CL↗

arXiv2025

AudioSet-R: A Refined AudioSet with Multi-Stage LLM Label Reannotation

AudioSet is a widely used benchmark in the audio research community and has significantly advanced various audio-related tasks. However, persistent issues with label accuracy and completeness remain critical bottlenecks that limit performance in downstream applications.To address the aforementioned challenges, we propose a three-stage reannotation framework that harnesses general-purpose audio-language foundation models to systematically improve the label quality of AudioSet. The framework employs a cross-modal prompting strategy, inspired by the concept of prompt chaining, wherein prompts are sequentially composed to execute subtasks (audio comprehension, label synthesis, and semantic alignment). Leveraging this framework, we construct a high-quality, structured relabeled version of AudioSet-R. Extensive experiments conducted on representative audio classification models--including AST, PANNs, SSAST, and AudioMAE--consistently demonstrate substantial performance improvements, thereby validating the generalizability and effectiveness of the proposed approach in enhancing label reliability.The code is publicly available at: https://github.com/colaudiolab/AudioSet-R.

cs.SD↗

arXiv2025

Signaling Human Intentions to Service Robots: Understanding the Use of Social Cues during In-Person Conversations

As social service robots become commonplace, it is essential for them to effectively interpret human signals, such as verbal, gesture, and eye gaze, when people need to focus on their primary tasks to minimize interruptions and distractions. Toward such a socially acceptable Human-Robot Interaction, we conducted a study ($N=24$) in an AR-simulated context of a coffee chat. Participants elicited social cues to signal intentions to an anthropomorphic, zoomorphic, grounded technical, or aerial technical robot waiter when they were speakers or listeners. Our findings reveal common patterns of social cues over intentions, the effects of robot morphology on social cue position and conversational role on social cue complexity, and users' rationale in choosing social cues. We offer insights into understanding social cues concerning perceptions of robots, cognitive load, and social context. Additionally, we discuss design considerations on approaching, social cue recognition, and response strategies for future service robots.

cs.HC↗