Searcharxiv⌕ Search

arXiv subjects

Yiming Zhao

Publications and source records attributed to Yiming Zhao.

At least 19 recordsLinked to original sources
arXiv2026

AlGaN/GaN Hall-Effect Sensor for In-Situ Magnetic Field Monitoring of the HSX Stellarator

Direct magnetic field sensors can address integration drift commonly observed in conventional inductive magnetic diagnostics used in fusion systems. In this work, an AlGaN/GaN Hall-effect sensor was fabricated, packaged, and deployed inside the Helically Symmetric eXperiment (HSX)---the first quasi-helically symmetric stellarator, operating with a 1 T on-axis magnetic field and up to 200 kW of launched electron cyclotron resonance heating (ECRH) power---for in-situ magnetic field monitoring near the plasma edge. The sensor leverages the high-mobility two-dimensional electron gas (2DEG) formed in the AlGaN/GaN heterostructure for sensitive magnetic field measurement, while the wide-bandgap GaN material system provides thermal robustness for harsh-environment operation. During 68 consecutive plasma discharge shots, the sensor remained functional and produced clear transient responses associated with plasma ignition and discharge dynamics. Comparisons between biased and unbiased operation, as well as plasma-discharge and coil-only shots, confirmed that the response originated from the biased Hall-effect sensor element. Furthermore, the sensor output exhibited temporal correlation with the plasma stored energy measured by the HSX diamagnetic loop across high-energy, late-breakdown, and failed-breakdown discharges.

physics.ins-det↗
arXiv2026

Decoupling Policy Extraction for Offline Reinforcement Learning

Offline RL methods commonly jointly train the actor and critic, where the critic is used to guide the actor toward higher-value actions. This coupled learning process is well motivated in online RL, where an improved actor collects new data that can further update the actor and the critic. However, training data remains fixed in offline RL, making actor-side policy improvement unable to generate new data to validate or correct the critic. Moreover, retaining this coupled paradigm leads to two related challenges. Firstly, actor updates can drift toward high-valued but potentially out-of-distribution (OOD) actions and amplify critic overestimation. Secondly, conservative value estimation or behavior-cloning regularization creates a difficult trade-off between suppressing OOD actions and selecting high-value actions within the data-supported region. Motivated by this observation, we revisit the conventional offline RL paradigm and propose decoupling policy improvement from actor training. Specifically, we train the actor solely to model the behavior distribution and perform policy improvement at inference time by reranking multiple actor-generated proposals with a separately learned critic. We refer to this paradigm as the decoupled policy extraction paradigm. Under such paradigm, the actor provides behavior-supported action candidates, while the critic performs value-based selection within this candidate set. Extensive experiments show that the decoupled policy extraction paradigm outperforms both behavior cloning and jointly learned offline RL methods, while remaining effective even with a naive Q-learning critic.

cs.LG↗
arXiv2026

FACET: Preserving Source Intent and Executable State in Terminal Task Synthesis

Training terminal agents requires scalable executable supervision, yet synthesizing high-quality terminal tasks remains challenging. Each task couples an instruction, an initialized environment, a reference solution, and an executable verifier; if these artifacts are generated from inconsistent assumptions, the resulting task may be unsolvable or incorrectly evaluated. Meanwhile, multi-stage synthesis can discard the goals, dependencies, state transitions, and procedural constraints encoded in the original sources. We present FACET (Fine-grained Agentic Construction of Executable Tasks), a framework that addresses both information preservation and cross-artifact consistency. FACET reconstructs related agent skills into coherent, information-rich scenarios, then realizes and repairs the execution environment before generating the final task artifacts. The resulting container state serves as shared grounding for the instruction, solution, and verifier, while execution-based validation and targeted repair correct artifact-specific failures without unnecessarily regenerating valid components. FACET produces complex terminal tasks with dense executable checks, and successful trajectories collected from these tasks provide effective, data-efficient supervision. Fine-tuning models across multiple scales consistently improves performance on Terminal-Bench 2.1, while analyses of alternative generation schemes support the importance of environment-grounded construction for task validity and solution-verifier alignment. These results establish source-intent preservation and shared executable-state grounding as key principles for scalable terminal-task synthesis.

cs.AI↗
arXiv2026

Sensing in Low-altitude Wireless Networks: Systems, Techniques, and Developments

The highly dynamic and safety-critical characteristics of low-altitude airspace render sensing an indispensable component of low-altitude wireless networks (LAWN). Although sensing techniques have been extensively studied under diverse paradigms, a prominent mismatch persists between state-of-the-art sensing schemes and the practical sensing demands of LAWN. To fill this research gap, this article systematically reviews LAWN-oriented sensing from the dimensions of system framework, core technologies, and research trends. Specifically, we first analyze the sensing system framework, covering concepts, services and tasks, nodes and targets, and scenarios for LAWN sensing. Next, we conduct a comparative analysis of existing sensing techniques from the perspectives of propagation medium, cooperation, methodology, and modality, analyzing their advantages and limitations. Then, we summarize promising future research directions for deployable LAWN sensing systems, covering non-cooperative and cooperative sensing, model-driven and data-driven sensing, and model-and-data-driven multi-modal sensing. Finally, we present a case study of a model-and-data-driven multi-modal method for real-time aerial target sensing. Compared with existing surveys on LAWN or sensing, this article delivers a more comprehensive, targeted review exclusively centered on LAWN sensing.

cs.NI↗
arXiv2026

Rethink Before You Execute: Adaptive Execution for World Action Models

World Action Models (WAMs) jointly predict future actions and the evolution of the environment. At each inference, a WAM generates a chunk of actions and the robot executes a fixed prefix before replanning. We argue that this fixed execution horizon is poorly matched to execution dynamics: the chunk reliability varies across task stages, so when to replan depends on the result of accumulated execution, not on the step counts. We propose TempoWAM (Timing Execution by Monitoring Progress Online), a lightweight plug-and-play execution scheme for WAMs. A Recurrent Progress Monitor first estimates task progress from the current observation, task instruction, remaining actions, and execution history; and an Adaptive Execution Protocol then evaluates whether the chunk is advancing the task to decide if replanning is needed. To bridge the training-deployment gap, the protocol is calibrated by a task-dependent calibration factor with online adaptation. Experiments on LIBERO, RoboTwin, and real-world tasks show that TempoWAM consistently improves the efficiency-success trade-off of WAM execution. On real robots, it reduces WAM inferences by 26.9% on easy tasks while maintaining success, and improves success by 13.3 points on difficult tasks.

cs.RO↗
arXiv2026

Video-DeepResearch: Towards the Next-Generation Multimodal Deepresearch Agent

We introduce Video-DeepResearch (Video-DR), extending multimodal agents from static images to continuous video streams, a setting that demands dense spatiotemporal grounding coupled with open-web exploration. Preliminary evaluations reveal two critical bottlenecks in current models: (1) modality bias, where agents bypass visual tools in favor of textual search, and (2) parametric knowledge leakage, where models rely on internal memory rather than genuine tool-augmented execution. To address these challenges, we propose Video-DR, featuring a decoupled perception-exploration pipeline with stage-wise tool unlocking that compels exhaustive cross-frame visual grounding prior to web retrieval. Our framework adopts a two-stage training recipe: supervised fine-tuning followed by Group Relative Policy Optimization (GRPO), enabling autonomous exploration that breaks the imitation-learning ceiling. Furthermore, we curate Video-DR-Bench, a human-AI collaborative benchmark comprising 200 complex, multi-hop VQA instances. Empirical results demonstrate that our Video-DeepResearch-35B-A3B establishes a new state-of-the-art of 64.0% average accuracy, surpassing proprietary Claude-4.5-Sonnet (59.0%) by 5.0 points and significantly outperforming GPT-5 (52.5%) and Gemini 2.5 Pro (57.5%). The 30B-A3B variant achieves 59.3%, competitive with Claude-4.5-Sonnet and demonstrating the effectiveness of our training paradigm even at compact scale. Code: https://github.com/Osilly/Vision-DeepResearch.

cs.CV↗
arXiv2026

Interleaved POMDP Planning for Multi-Object Search in Unknown Multi-Room Household Environments

Multi-object search in unknown household environments requires planning under extensive uncertainty - from unknown object locations to cluttered spaces with unobserved obstacles. POMDPs offer a principled framework for such problems but remain intractable in large domains. We propose Inter-POMDP, a novel interleaved POMDP planning algorithm that decomposes this challenge into two interacting levels: a high-level POUCT planner reasons over object distributions using LLM-informed histogram beliefs, while a low-level motion planner models navigation uncertainty with obstacle-aware particle beliefs as domain knowledge to guide high-level POUCT. This interleaved design balances planning quality and efficiency despite the large search space across unknown multi-room environments. Both simulation and real-world experiments show that our Inter-POMDP algorithm reduces collision counts by up to 63%, navigation steps by up to 35%, and detection counts by up to 32% compared with baseline methods. Full videos are https://sites.google.com/view/inter-pomdp

cs.RO↗
arXiv2026

TSD: A Physics-Inspired Trajectory Saliency Detector for Efficient Imitation Learning

For imitation learning in robotic manipulation, high data collection costs result in the scarcity of high quality data. In this paper, we leverage the inherent heterogeneity of trajectories to address this challenge. Based on our observations of manipulation tasks, we categorize motions into transitional, precise, and agile types, defining the latter two as trajectory saliency due to their criticality to task success in contrast to the prevalent but less relevant transitional motions. Therefore, we propose the Trajectory Saliency Detector (TSD), a training-free and plug-and-play framework to identify trajectory saliency. TSD employs two physically-grounded metrics: spatial entropy to capture fine-grained manipulation and centripetal acceleration to detect agile maneuvering. We further leverage TSD to develop a dataset compression method that reduces training costs and a dataset expansion strategy that improves data collection efficiency. Extensive experiments in both simulation and real-world settings demonstrate that models trained on TSD-condensed datasets achieve comparable or even superior performance with 25% less data on average. These results validate the effectiveness of our dataset compression and expansion strategies, thereby confirming the utility of TSD. Consequently, TSD offers a scalable and cost-effective pathway to synthesize information-dense datasets for efficient robot learning. Project page: https://trajectory-saliency-detector.github.io/trajectory-saliency-detector/

cs.RO↗
arXiv2026

ReFPO: Reflow Regularization for Flow Matching Policy Gradients

We present Reflow-regularized Flow Matching Policy Gradients (ReFPO), a simple online RL method that adds explicit Reflow regularization to FPO for efficient flow-based control. We uncover a key structural property: the gradient updates in Flow Matching Policy Gradients (FPO) can be interpreted as an implicit advantage-weighted Reflow process, providing a new geometric perspective on flow-based policy gradients. Building on this insight, ReFPO introduces an explicit geometric regularizer that can be implemented with a single line of code change without incurring additional computational overhead or auxiliary distillation stages. By synergizing advantage-guided updates with path rectification, our method reduces CFM proxy-ratio spikes, stabilizes PPO-style training, and enables high-fidelity one-step inference that often matches or exceeds multi-step performance. We experimentally demonstrate that ReFPO improves average performance and discretization robustness across GridWorld, MuJoCo Playground, and high-dimensional Humanoid Control tasks, providing a scalable and stable approach for generative policies in complex physical simulations.

cs.RO↗
arXiv2026

MimicIK: Real-Time Generative Inverse Kinematics from Teleoperation with FK Consistency

Inverse kinematics (IK) remains a critical bottleneck for real-time robot manipulation. Classical numerical solvers achieve high geometric precision but often suffer from discontinuous branch switching and unstable behavior near kinematic singularities during closed-loop deployment. Meanwhile, learned IK approaches frequently struggle to balance spatial accuracy, motion smoothness, and real-time efficiency, particularly when trained on noisy human teleoperation data. We present \textbf{MimicIK}, a real-time generative inverse kinematics framework that learns smooth and robust joint-space motion priors from teleoperation demonstrations through conditional flow matching. Given the current joint configuration and a target end-effector pose, MimicIK predicts continuous delta-joint commands using an efficient two-step iterative refinement process based on a Minimal Iterative Policy (MIP) backbone. To enforce physical consistency, we further introduce an FK consistency loss, a differentiable forward-kinematics regularization that penalizes task-space deviations from the target pose during training. We evaluate MimicIK on a real-world 6-DOF robot dataset containing 8,848 teleoperation demonstrations. MimicIK achieves a mean position error of 4.65 mm, a 10 mm success rate of 92.01\%, and a trajectory spike rate of only 7.99\%. Compared with a UNet diffusion baseline, our method improves both spatial accuracy and motion smoothness while reducing inference latency from 21.66 ms to 6.74 ms. Furthermore, unlike deterministic MLP baselines that catastrophically diverge under out-of-distribution deployment, MimicIK remains stable near singular configurations and enables robust 20 Hz real-time control on deployment hardware.

cs.RO↗
arXiv2026

Acting While Understanding: Asynchronous Semantic-Action Decoupling for Real-Time Vision-Language-Action Models

Vision-Language-Action models (VLAs) have demonstrated strong task understanding and generalization in robotic manipulation, yet the high computational cost of full-model inference limits their deployment in low-latency, high-frequency closed-loop control. We propose an asynchronous semantic-action decoupling framework that separates semantic understanding from action generation along the internal semantic-action interface of existing VLAs, without redesigning the vision-language backbone or introducing an external planner. A low-frequency understanding module asynchronously updates reusable semantic conditions, while a high-frequency action module continuously outputs control actions without repeatedly invoking the full model. To mitigate the temporal mismatch between stale semantics and the current execution state, we further introduce historical action conditioning and time-misalignment training, which provide short-horizon execution context and improve feedback control robustness under stale semantic conditions. Experiments on LIBERO with $\pi_{0.5}$ and UniVLA, together with real-robot deployment using UniVLA, show that the proposed framework achieves up to 35.6 Hz server-side action-module inference throughput and offers a low-intrusion path to high-frequency closed-loop control without running full VLA inference at control rate.

cs.RO↗
arXiv2026

Elastic Queries Reinforcement Learning: Self-Aware Policy Execution for VLA Models

Vision-language-action (VLA) models are powerful action generators for robot manipulation, but they are typically executed with fixed inference and replanning schedules. This rigidity ignores the uneven difficulty of robot control: contact-rich or uncertain states may need more computation and fresher feedback, while easier states can often be handled with fewer inference steps and longer open-loop execution. We propose Elastic Queries Reinforcement Learning (EQRL), a framework that makes each VLA policy query elastic. A lightweight latent-schedule adaptor jointly selects the latent input, denoising budget, and action chunk length, without fine-tuning the underlying VLA model. To make scheduling difficulty-aware, EQRL trains a critic over the joint latent-schedule action and derives a state difficulty signal from critic ensemble disagreement. This signal guides compute toward difficult states, while a learned residual allows task-driven correction. We formulate variable chunk execution as query-level macro-action RL with chunk-dependent discounting and an amortized number-of-function-evaluations (NFE) budget. Across simulation and real-robot manipulation, EQRL reduces amortized inference cost while preserving or improving task success.

cs.RO↗
arXiv2026

AnimaSpark: A Feed-Forward Method for Animating Arbitrary 3D Objects

While recent advancements in generative AI have substantially accelerated static 3D model creation workflows, the synthesis of category-agnostic 3D animations remains a significant bottleneck in 3D asset production. Current methods for category-agnostic animation generation exhibit critical limitations in inference speed, motion quality, and adherence to textual prompts, thereby leaving the process dependent on labor-intensive manual artistry. To address these challenges, this paper introduces AnimaSpark, a novel pipeline for category-agnostic 3D animation generation. Our approach is motivated by the key insight that for many fundamental motions in the 3D world, the corresponding joint transformations can often be effectively modeled within a two-dimensional subspace. The pipeline begins by rendering a rigged static 3D model into multi-layered image representations of its mesh and skeleton, which are subsequently fed into a video generation model. We then employ a keypoint tracking algorithm on the generated video to capture the motion of the skeletal joints projected onto the camera's viewing plane. In the final stage, we distill the planar translations and rotations from these tracked keypoints and lift them from the 2D domain into 3D space to animate the character. Comprehensive evaluations reveal that our method achieves superior performance over existing state-of-the-art techniques across key metrics, including text-motion alignment, quality of motion, and computational efficiency.

cs.CV↗
arXiv2026

ConSteer-RL: Steering Reasoning Capabilities in Large Language Models via Confidence-Aware Reinforcement Learning

Reinforcement Learning from Verifiable Rewards (RLVR) has recently become a key paradigm for improving the reasoning abilities of Large Language Models (LLMs), yet it remains limited by sparse binary rewards and its ignorance of model-internal uncertainty. In this paper, we propose ConSteer-RL, a simple yet effective framework that integrates token-level confidence signals derived from model log-probabilities into RLVR training. Specifically, building upon the Group Relative Policy Optimization (GRPO) framework, we construct a confidence-aware reward by aggregating per-token probabilities into a scalar confidence score and incorporating it into an awareness-based reward shaping mechanism that penalizes overconfident errors while reinforcing correct and confident reasoning. Experimental results demonstrate that ConSteer-RL consistently outperforms strong GRPO baselines, achieving average improvements of 2.3%-4.0% across different model scales.

cs.LG↗
arXiv2026

Boundary-Protection W8A8 HiFloat8 Quantization for Large-Scale Text-to-Video Diffusion Transformers

We present a post-training quantization (PTQ) approach for Wan2.1-T2V-14B, a 14-billion-parameter text-to-video diffusion transformer, targeting the W8A8 HiFloat8 (HiF8) format on Ascend 910B NPUs. A central challenge in quantizing video DiT models is the heterogeneous activation distribution across transformer blocks: boundary blocks (the first and last few blocks) exhibit fundamentally different statistical properties from middle blocks, making uniform quantization ineffective. We conduct a systematic per-block activation analysis across all 40 WanAttentionBlocks and use the findings to motivate a boundary-protection strategy that retains the first two and last three blocks in BF16 while quantizing the remaining 35 blocks with W8A8 HiF8. The proposed PTQ method matches or marginally exceeds the BF16 baseline on all five VBench dimensions evaluated, indicating no measurable accuracy loss within the 5-prompt evaluation set. An ablation study over four protection configurations confirms that full boundary protection yields the highest average VBench score, validating the data-driven block selection. We additionally investigate quantization-aware training (QAT) as a complementary fine-tuning stage and analyze the conditions under which it fails to outperform plain PTQ on single-card hardware.

cs.CV↗
arXiv2026

ACC: Compiling Agent Trajectories for Long-Context Training

Recent development of agents has renewed demand for long-context reasoning capacity of LLMs. However, training LLMs for this capacity requires costly long-document curation or heuristic context synthesis. We observe that agents produce massive trajectories when solving problems, invoking tools and receiving environment observations across many turns. The evidence needed to answer the original question is thus scattered throughout these turns, requiring integration of distant context segments. Nevertheless, standard agent SFT masks tool responses and only trains turn-level tool selection, creating a supervision blind spot where these scattered signals go unused. We propose Agent Context Compilation (ACC), which converts trajectories from search, software engineering, and database querying agents into long-context QA pairs that combine the original question with tool responses and environment observations gathered across multiple turns, training the model to answer directly without tool use. This makes the dependencies between the question and the evidence explicit, enabling direct supervision of long-context reasoning over distant segments without additional annotation. ACC is a simple but effective approach that can be combined with any existing long-context extension or training method, providing scalable supervised fine-tuning data. We validate ACC on long-range dependency modeling tasks through MRCR and GraphWalks, challenging benchmarks requiring cross-turn coreference resolution and graph traversal over extended contexts. Training Qwen3-30B-A3B with ACC achieves 68.3 on MRCR (+18.1) and 77.5 on GraphWalks (+7.6), results comparable to Qwen3-235B-A22B, while preserving general capabilities on GPQA, MMLU-Pro, AIME, and IFEval. Further mechanism analysis reveals that the ACC-trained model exhibits task-adaptive attention restructuring and expert specialization.

cs.CL↗
arXiv2026

SaaSBench: Exploring the Boundaries of Coding Agents in Long-Horizon Enterprise SaaS Engineering

As autonomous coding agents become capable of handling increasingly long-horizon tasks, they have gradually demonstrated the potential to complete end-to-end software development. Although existing benchmarks have recently evolved from localized code editing to from-scratch project generation, they remain confined to structurally simplified, single-stack applications. Consequently, they fail to capture the heterogeneous environments, full-stack orchestration, and system-level complexity of real enterprise Software as a Service (SaaS) systems, leaving a critical gap in assessing agents under realistic engineering constraints. To fill this gap, we introduce SaaSBench, the first benchmark designed to explore the boundaries of AI agents in enterprise SaaS engineering. Spanning 30 complex tasks across 6 SaaS domains with 5,370 validation nodes, it incorporates 8 programming languages, 6 databases, and 13 frameworks to meticulously mirror real-world software heterogeneity. Furthermore, we design a dependency-aware hybrid evaluation paradigm tailored for complex systems with long horizons and multi-component coupling, enabling fine-grained, reproducible assessment. Crucially, our extensive experiments reveal a striking insight: the primary bottleneck for state-of-the-art agents is not generating isolated code logic, but successfully configuring and integrating a multi-component system. Over 95\% of task failures occur before agents even reach deep business logic, with models often falling victim to overconfidence and prematurely halting during foundational system setup, or getting trapped in ineffective debugging loops. We hope SaaSBench serves as a practical and challenging testbed to drive the evolution of reliable, system-level coding agents. The code is available at \url{https://github.com/ShadeCloak/SaaSbench}.

cs.SE↗
arXiv2026

VideoSeeker: Incentivizing Instance-level Video Understanding via Native Agentic Tool Invocation

Large Vision-Language Models (LVLMs) have shown significant progress in video understanding, yet they face substantial challenges in tasks requiring precise spatiotemporal localization at the instance level. Existing methods primarily rely on text prompts for human-model interaction, but these prompts struggle to provide precise spatial and temporal references, resulting in poor user experience. Furthermore, current approaches typically decouple visual perception from language reasoning, centering reasoning around language rather than visual content, which limits the model's ability to proactively perceive fine-grained visual evidence. To address these challenges, we propose VideoSeeker, a novel paradigm for instance-level video understanding through visual prompts. VideoSeeker seamlessly integrates agentic reasoning with instance-level video understanding tasks, enabling the model to proactively perceive and retrieve relevant video segments on demand. We construct a four-stage fully automated data synthesis pipeline to efficiently generate large-scale, high-quality instance-level video data. We internalize tool-calling and proactive perception capabilities into the model via cold-start supervision and RL training, building a powerful video understanding model. Experiments demonstrate that our model achieves an average improvement of +13.7% over baselines on instance-level video understanding tasks, surpassing powerful closed-source models such as GPT-4o and Gemini-2.5-Pro, while also showing effective transferability on general video understanding benchmarks. The relevant datasets and code will be released publicly.

cs.CV↗