arXiv subjects

Liang Wang

Publications and source records attributed to Liang Wang.

At least 19 recordsLinked to original sources

arXiv2026

AWARE: Action-Entropy-Weighted and Attention-Guided Retrieval of Observation Evidence for Vision-Language-Action Models

Vision-Language-Action (VLA) models combine multimodal perception with robotic control, demonstrating remarkable potential for generalizable manipulation. Yet, enhancing their execution typically relies on auxiliary sensory inputs (e.g., depth, 3D point clouds) or costly fine-tuning. Inspired by the insight that Feed-Forward Networks (FFNs) act as key-value memory, we propose Action-Entropy-Weighted and Attention-Guided Retrieval of Observation Evidence (AWARE), an effective, training-free, and plug-and-play module for VLAs. Specifically, AWARE tracks layer-wise uncertainty via an element-wise Action Entropy metric, and retrieves observation evidence into the element's FFN through associative memory lookup. Crucially, the retrieved features are weighted by instruction-guided cross-modal attention to isolate task-relevant observation cues, while injection strength is adaptively scaled by the element's entropy and relative network depth. This selectively channels observation evidence to high-entropy action dimensions and deep layers where grounding decays, while leaving converged representations intact. Across LIBERO, SIMPLER, CALVIN, and real-world robot manipulation, AWARE consistently improves diverse VLA architectures with minimal computational overhead, achieving the largest gains on long-horizon tasks without architectural changes or additional training. Project page: https://jiabingyang01.github.io/AWARE/.

cs.CV↗

arXiv2026

Coupling Execution Horizon and Sampling Depth in Flow-Matching VLA Policies

Flow-matching Vision-Language-Action (VLA) models generate expressive continuous trajectories for robotic manipulation, yet balancing real-time responsiveness with execution stability remains a persistent challenge. Standard deployment workflows typically treat numerical solver depth and execution horizon as decoupled choices. The horizon amortizes inference latency, while solver depth is kept high to preserve integration fidelity. In this paper, we reveal a fundamental compute-execution coupling in flow-matching VLAs governed by a systematic sign reversal. While coarse numerical integration degrades precision under rapid replanning, it markedly improves task performance under extended open-loop commitment. Geometrically, numerical truncation contracts lateral action dispersion around an invariant trajectory centroid, serving as an implicit regularizer that suppresses compounding open-loop drift. Across LIBERO, RoboTwin 2.0, and real-world Franka manipulation on representative VLA backbones including $π_0$, $π_{0.5}$, and GR00T, exploiting this joint coupling consistently achieves superior task success while drastically lowering inference latency, entirely training-free and without architectural modifications. These findings establish numerical sampling depth as an operational variance regulator, shifting its role from a static precision budget to an active deployment knob in robot learning. Project page: https://jiabingyang01.github.io/Horizon-Depth-VLA/

cs.RO↗

arXiv2026

Deep-Learning Identification and Global Characterization of Energetic O2+ Beams at Mars

Martian ionospheric ions can be accelerated away from Mars by the solar wind, contributing substantially to atmospheric ion escape. Here, we develop a U-Net-based image segmentation model to identify a distinct population of escaping planetary ions, energetic O2+ beams, from a decade of Mars Atmosphere and Volatile EvolutioN (MAVEN) observations and perform an event-based statistical analysis of their global properties and dependence on external conditions. We find that O2+ beams are widely distributed throughout the Martian plasma environment, with a strong preference for the +E hemisphere, where the solar wind convective electric field points away from Mars. This pronounced spatial asymmetry indicates that the convective electric field plays a major role in beam formation. In the solar wind and magnetosheath, these beams are consistent with the plume. Their occurrence-weighted flux agrees with previous long-term plume estimates, whereas the flux during identified beam intervals is substantially higher. This highlights the advantage of our event-based approach in capturing the intrinsic properties of O2+ plumes. In the induced magnetotail, beam formation remains strongly associated with the convective electric field, while the Hall electric field can provide additional energization. The beam properties are primarily controlled by upstream solar wind conditions, with no apparent influence from solar extreme ultraviolet irradiance. These results might improve our understanding of ion escape at Mars. More broadly, we also demonstrate the potential of deep learning for identifying transient plasma populations from spacecraft observations.

physics.space-ph↗

arXiv2026

Relational Geometry Forcing for World Action Models

World Action Models (WAMs) have shown strong potential for robot manipulation by jointly learning visual dynamics and executable actions. However, their video and action objectives do not explicitly constrain the 3D geometry that is crucial for precise manipulation. Recently, geometry-aware WAMs introduce depth or 4D prediction branches, whereas representation alignment methods supervise intermediate WAM features with pretrained 3D foundation models. Yet we find that such pointwise alignment benefits the video features but degrades action learning, since dense geometric feature targets overly restrict compact action features. To address this limitation, we propose Relational Geometry Forcing (RGF), which transfers geometric priors from a frozen 3D foundation model by aligning pairwise feature relations instead of pointwise features. Specifically, RGF applies three complementary relational objectives, where intra-frame and cross-view relations transfer scene geometry and multi-view correspondence to the video features, while action-temporal relations couple action dynamics to geometric scene evolution. Applied only during training, RGF requires no explicit geometry branch or extra inference module. Extensive experiments in simulation and the real world demonstrate the effectiveness of Relational Geometry Forcing, showing consistent gains across diverse manipulation scenarios and WAM paradigms.

cs.RO↗

arXiv2026

When Seeing Is Not Believing -- A Benchmark for Search-Grounded Video Misinformation Detection

Video misinformation increasingly operates at the semantic and evidential level: authentic footage may be selectively edited, temporally reordered, spliced across sources, or augmented with AI-generated content to construct false narratives. Such evidence-dependent manipulations cannot be reliably verified from the input video alone, because the missing, reordered, replaced, or recontextualized evidence lies outside the video itself. We introduce \textbf{EVID-Bench}, a benchmark for search-grounded video misinformation detection, where a system must search the open web for related videos and identify what information is false through cross-video comparison. EVID-Bench comprises 222 videos spanning 9 manipulation types across 3 categories: AI generation, single-source editing, and multi-source editing. All samples are verified to be undetectable by frontier models through visual inspection alone. We evaluate nine frontier multimodal models using a retrieval-augmented verification baseline. The best system achieves only 61.43\% point-level accuracy and 43.24\% video-level accuracy, while AI-generated manipulations remain especially challenging. Error analysis reveals recurring challenges: models fixate on irrelevant anchors, misattribute synthetic content to editorial splicing, and terminate search prematurely before fully explaining the manipulation.

cs.CV↗

arXiv2026

Noncommutative maximal weak type $(1,1)$ estimates for convolution operators under $L_1$-Hörmander conditions

We prove weak type $(1,1)$ estimates for families of scalar convolution operators acting on operator-valued function spaces. A joint $L_1$-Hörmander condition, together with a maximal strong type estimate at one exponent $1<r\leq\infty$, suffices. The kernels may be complex-valued and need not have uniformly bounded $L_1$ norms. Under a size condition, the result also yields a weak endpoint estimate for all sharp truncation radii of singular kernels. This resolves, for convolution kernels, the problem mentioned in \cite[Remark~5.5]{parcet09pseudo} and recalled in Parcet's recent survey \cite[Remark~1.3]{Parcet2026ICM}. The proof involves not only a simultaneous dyadic representation of the operators but also a noncommutative maximal weak type $(1,1)$ estimates of the resulting dyadic operators. This approach seems entirely new even in the commutative setting. Numerous applications are provided. On the one hand, it yields various maximal inequalities and pointwise convergence results for integrable kernels and dyadic approximations of the identity, monotone $L\log L$ majorants, and Poisson integrals on nilpotent groups and symmetric spaces. On the other hand, via various transference techniques, it implies maximal inequalities for Fourier multipliers, Schur multipliers and double operator integrals, together with bilateral almost uniform convergence of dyadic Fourier summation methods on $L_1$ of group von Neumann algebras.

math.FA↗

arXiv2026

A noncommutative weak $L_1$-estimate for rough singular integrals

In this paper, we establish a noncommutative version of Andreas Seeger's weak type $(1,1)$ estimate for singular integral operators with kernels satisfying angular $L\log L$ integrability and an averaged radial Dini condition. The strategy for obtaining the sharp integrability condition involves not only some new estimates of the off-diagonal good and bad functions in Parcet's noncommutative Calderón-Zygmund decomposition but also certain essential modifications of Seeger's kernel decomposition and estimates.

math.OA↗

arXiv2026

A Mizohata--Takeuchi-type estimate for the hyperbolic paraboloid

We prove Mizohata--Takeuchi-type estimates for the Fourier extension operator associated with the truncated hyperbolic paraboloid in $\R^3$. Our main result controls the weighted $L^2$ norm of the extension operator by the $L^2$ mass of the weight over a multiscale family of parallelepipeds arising from $\ell^2$ decoupling for the hyperbolic paraboloid. We deduce this result from a weighted $L^2$ estimate for functions whose Fourier transforms are supported in a small neighborhood of the hyperbolic paraboloid, which we establish using a bilinear refined decoupling estimate.

math.CA↗

arXiv2026

What Matters for World Modeling in Vision-Language-Action Models?

Integrating world models into Vision-Language-Action models (VLAs) has emerged as a promising direction, enabling policies to anticipate future dynamics beyond reactive observation-to-action mapping. However, existing approaches vary simultaneously in their world representations, integration paradigms, and policy backbones, making it difficult to determine what truly matters for effective world model integration. To disentangle these factors, we categorize existing efforts into four representative paradigms, namely Implicit Future Alignment, Explicit Future Query, MoT Future Expert, and External World Predictor. We instantiate each within a unified framework and systematically evaluate these paradigms with diverse world representations and policy backbones across in-domain, out-of-domain, long-horizon, dynamic, few-shot, and real-world settings. Our study yields four main findings: 1) world representation effectiveness is paradigm-dependent, whereas 2D semantic representations offer robust, cross-paradigm gains; 2) no single paradigm is universally optimal, yet the decoupled MoT Future Expert offers competitive overall performance, particularly in OOD robustness; 3) world modeling consistently benefits policies built on standard VLMs, while decoupled designs exhibit greater compatibility with pretrained VLAs; and 4) world modeling improves performance over action-only baselines across multiple settings, with notable gains on long-horizon and dynamic tasks. Project page and video demonstrations are available at https://dylanmqs.github.io/WM-in-VLAs/.

cs.RO↗

arXiv2026

When Non-Adversarial Noise Hurts Learning: Benchmarking Skill Self-Evolution in Agents

Large language model agents increasingly improve by distilling experience into reusable skills, but errors in that experience can persist in learned rules and affect later tasks. Existing evaluations largely examine clean learning or perturbations encountered after skills are frozen, leaving robustness during self-evolution insufficiently understood. We introduce RSE-Bench, to our knowledge the first benchmark for evaluating skill self-evolution under ordinary, non-adversarial noise. It spans four domains and six benchmarks, with controlled perturbations to instructions, environmental evidence, execution records, and outcome--record pairings that preserve the underlying tasks and native scoring rules. The protocol separates noise during learning from input-side noise at evaluation, testing frozen skill stores on held-out tasks. We also introduce \textsc{RSEClaw}, a method-agnostic framework supporting consistent noise injection and process diagnostics of actual exposure and skill updates. Across three methods and two backbones, noisy learning retains only $25.5\%$--$42.1\%$ of the aggregate gain achieved by clean self-evolution, depending on the noise type. Perturbations confined to execution records and outcome--record pairings cause average losses comparable to input-side noise, despite leaving the original executions and evaluations intact. Diagnostics further identify differences in candidate skill text that coexist with small changes in mean held-out performance.

cs.LG↗

arXiv2026

Open-MMUnlearning: Unifying Methods and Evaluation for MLLM Unlearning

As multimodal large language models (MLLMs) become more capable and widely deployed, concerns about privacy and safety have become increasingly pressing. Machine unlearning offers one approach to addressing these concerns by removing designated information from trained models while preserving unrelated capabilities. However, fragmented implementations and evaluation protocols, incomplete robustness testing, and limited understanding of metric reliability make progress in MLLM unlearning difficult to assess systematically. We introduce Open-MMUnlearning, an open-source, extensible framework that integrates target-model preparation, multimodal data processing, unlearning, and evaluation through shared interfaces and structured configurations. The framework supports five benchmarks spanning privacy, safety, and copyright, eight MLLMs from four model families, and twelve unlearning methods. Its evaluation suite jointly assesses forgetting effectiveness, retained utility, and robustness to model interventions, adversarial inputs, and membership inference attacks. Using a common evaluation protocol, we compare ten representative unlearning methods. In this comparison, GD and MIP-Editor tie for the highest overall score: GD achieves the highest Forget Quality, while MIP-Editor preserves more Model Utility. We further introduce a metric meta-evaluation protocol that tests faithfulness using models with controlled exposure to target knowledge and robustness under quantization and relearning. Among the thirteen evaluated metrics, BLEU achieves the highest aggregate reliability score. KS-Test attains the highest faithfulness AUC but performs less well on robustness. Together, the framework and these findings support reproducible comparison of MLLM unlearning methods and systematic assessment of evaluation reliability.

cs.AI↗

arXiv2026

PIGEONS: Incremental Long-Video Captioning with Persistent Identity Gallery and Evolving Overarching Narrative Summary

Detailed video captioning has become an essential semantic interface for multi-modal understanding and generation. While existing methods mainly target short videos, extending them to long videos remains challenging, primarily due to limited context windows of current video understanding models. A naive solution is to adopt segment-wise captioning followed by concatenation, yet this paradigm often suffers from inconsistent subject references and fragmented long-range narratives. To mitigate these issues, we introduce PIGEONS, an incremental long-video captioning model built upon a Persistent Identity Gallery and an Evolving Overarching Narrative Summary. Before caption generation, PIGEONS scans the entire video to construct an identity gallery that captures key subjects along with their visual references, thereby providing a stable referential anchor. The model then performs clip-by-clip captioning conditioned on the persistent gallery and an evolving overarching narrative summary, which compactly maintains and updates essential long-range contextual information after each captioning step. To enable these capabilities, we develop a three-stage training strategy that integrates supervised fine-tuning and reinforcement learning. Extensive experiments demonstrate that PIGEONS markedly improves long-video captioning performance, achieving superior accuracy, completeness, and subject referential consistency over existing methods, while maintaining competitive performance on short-video captioning.

cs.CV↗

arXiv2026

Hand2Robot: Object-Centric 4D Hand Reconstruction from Human Demonstrations for Robotics

Human demonstrations can capture task-relevant manipulation without using the target robot, but the recorded hand motions cannot directly serve as robot action supervision. Hand reconstruction and motion retargeting provide a direct route to such supervision, yet recovering hand motion alone does not ensure physically consistent hand-object interactions. To address this issue, we introduce Hand2Robot, an object-centric 4D hand reconstruction framework for robot policy post-training. Given videos of humans performing the target task, Hand2Robot extracts interaction clips, reconstructs object geometry, and generates robot action supervision through hand-object interaction optimization and motion retargeting. For object tracking, we combine observation confidence with temporal motion constraints to select a globally consistent object pose trajectory from per-frame pose candidates. For hand refinement, we use object geometry and motion as a reference and combine staged 2D alignment with object-local contact constraints to refine hand trajectories and finger joint configurations. Experiments on DexYCB and HO3D-v3 show improved object pose accuracy and hand-object interaction quality. Experiments on three real-robot dexterous manipulation tasks further show that reconstructed and retargeted human demonstrations can effectively supplement limited robot data for target-task policy post-training.

cs.RO↗

arXiv2026

Frequency-Guided Diffusion Model with Perturbation Training for Skeleton-Based Video Anomaly Detection

Video anomaly detection (VAD) is a vital yet complex open-set task in computer vision, commonly tackled through reconstruction-based methods. However, these methods struggle with two key limitations: (1) insufficient robustness in open-set scenarios, where unseen normal motions are frequently misclassified as anomalies, and (2) an overemphasis on, but restricted capacity for, local motion reconstruction, which are inherently difficult to capture accurately due to their diversity. To overcome these challenges, we introduce a novel frequency-guided diffusion model with perturbation training. First, we enhance robustness by training a generator to produce perturbed samples, which are similar to normal samples and target the weakness of the reconstruction model. This training paradigm expands the reconstruction domain of the model, improving its generalization to unseen normal motions. Second, to address the overemphasis on motion details, we employ the 2D Discrete Cosine Transform (DCT) to separate high-frequency (local) and low-frequency (global) motion components. By guiding the diffusion model with observed high-frequency information, we prioritize the reconstruction of low-frequency components, enabling more accurate and robust anomaly detection. Extensive experiments on five widely used VAD datasets demonstrate that our approach surpasses state-of-the-art methods, underscoring its effectiveness in open-set scenarios and diverse motion contexts. Our project website is https://xiaofeng-tan.github.io/projects/FG-Diff.

cs.CV↗

arXiv2026

Point-Supervised Visual Learning for GNSS Measurement Quality Assessment

Multipath and non-line-of-sight (NLOS) propagation degrade global navigation satellite system (GNSS) positioning accuracy in urban environments. Although fisheye sky segmentation identifies obstructions along satellite directions, binary visibility does not fully characterize observation quality. This study proposes a point-supervised visual framework that predicts GNSS observation faults from fisheye images using supervision derived from GNSS observations. During offline labeling, a vision-assisted differential GNSS procedure selects multiple reference satellites based on sky visibility and elevation constraints. Double-difference pseudorange residuals automatically generate fault labels and sample importance weights at satellite projection locations. A residual-weighted loss is computed only at valid labeled points, enabling learning from sparse supervision without manual pixelwise fault annotations. The trained network generates dense fault probability maps, which are sampled at satellite projection locations to weight observations in weighted least squares positioning, reducing the influence of potentially faulty observations. The framework is evaluated using U-Net, DeepLabv3+, DINOv3, and SAM3 on urban vehicle data collected on a different day from the training data. All four models reduce horizontal root mean square error (RMSE) by 29.51% to 36.07% compared with SAM binary sky segmentation, with U-Net achieving the lowest RMSE of 0.78 m. The framework also provides positioning solutions during some intervals when the segmentation baseline fails. These results demonstrate the framework's generalization ability and show that GNSS-derived point supervision enables fisheye images to provide observation quality information beyond binary sky visibility, supporting GNSS fault detection.

eess.SP↗

arXiv2026

MotionRFT: Unified Reinforcement Fine-Tuning for Text-to-Motion Generation

Text-to-motion generation has advanced with diffusion- and flow-based generative models, yet supervised pre-training remains insufficient to align models with high-level objectives such as semantic consistency, realism, and human preference. Existing post-training methods have key limitations: they (1) target a specific motion representation, such as joints, (2) optimize a particular aspect, such as text-motion alignment, and may compromise other factors; and (3) incur substantial computational overhead, data dependence, and coarse-grained optimization. We present a reinforcement fine-tuning framework that comprises a heterogeneous-representation, multi-dimensional reward model, MotionReward, and an efficient, fine-grained fine-tuning method, EasyTune. To obtain a unified semantics representation, MotionReward maps heterogeneous motions into a shared semantic space anchored by text, enabling multi-dimensional reward learning; Self-refinement Preference Learning further enhances semantics without additional annotations. For efficient and effective fine-tuning, we identify the recursive gradient dependence across denoising steps as the key bottleneck, and propose EasyTune, which optimizes step-wise rather than over the full trajectory, yielding dense, fine-grained, and memory-efficient updates. Extensive experiments validate the effectiveness of our framework, achieving FID 0.132 at 22.10 GB peak memory for MLD model and saving up to 15.22 GB over DRaFT. It reduces FID by 22.9% on joint-based ACMDM, and achieves a 12.1% R-Precision gain and 21.7% FID improvement on rotation-based HY-Motion. Our project page with code is publicly available.

cs.CV↗

arXiv2026

EgoPhys: Estimating Peak Contact Force and Mechanical Work from Egocentric Manipulation Video

Physically grounded manipulation of articulated objects requires understanding both the maximum forces encountered during contact and the work performed as their parts move. Peak contact force and mechanical work quantify these complementary aspects, but estimating them from egocentric video is challenging because physical interaction cues are local and indirect. Moreover, peak force is associated with brief contact events, whereas mechanical work depends on force-motion coupling throughout the contact duration. To address these challenges, we propose EgoPhys, an RGB-only framework comprising Contact-Aware Spatial Aggregation (CASA) and Target-Specific Multi-Expert Temporal Routing (TMTR). CASA integrates appearance and geometry features to emphasize interaction-relevant cues, while TMTR models semantic, event, and motion cues with specialized temporal experts and routes them separately for force and work prediction. On the test split from Hoi! dataset, EgoPhys substantially improves predictions of peak force and mechanical work, achieving MAEs of $5.205 \pm 0.584$ $N$ and $0.894 \pm 0.081$ $J$, respectively.

cs.CV↗

arXiv2026

ViDAL: A Visual Dynamics-Grounded Action Latent Space for Vision-Language-Action Models

Vision-Language-Action (VLA) models have become a central paradigm for robot policy learning, which predict actions in three forms: raw action chunks, discrete action tokens, or continuous action latents. However, existing action representations primarily model action trajectories, with limited consideration of the visual dynamics induced by these actions. We introduce ViDAL, a Visual Dynamics-grounded Action Latent Space that anchors continuous action latents in the future visual dynamics of the scene. Specifically, ViDAL learns action latent space by training an Action Variational Autoencoder (Action VAE) to reconstruct action chunks while aligning its latent with future scene dynamics. When integrated into downstream robot policies, the proposed Action VAE serves as a plug-in action interface compatible with multiple VLA architectures and enables optional future-video prediction as an additional capability. Empirically, ViDAL outperforms competitive baselines on LIBERO with 98.1% average success, improves a multi-task $π_{0.5}$ policy on RoboTwin 2.0 from 54.3% to 65.5% (Clean) and from 33.2% to 43.1% (Random) success rates over 50 dual-arm tasks, and yields 20.0% and 23.4% absolute success-rate gains on real-world single-arm Franka and dual-arm ARX robot platforms.

cs.RO↗