arXiv subjects

Hao Luo

Publications and source records attributed to Hao Luo.

At least 55 records · Page 3Linked to original sources

arXiv2026

Joint-Aligned Latent Action: Towards Scalable VLA Pretraining in the Wild

Despite progress, Vision-Language-Action models (VLAs) are limited by a scarcity of large-scale, diverse robot data. While human manipulation videos offer a rich alternative, existing methods are forced to choose between small, precisely-labeled datasets and vast in-the-wild footage with unreliable hand tracking labels. We present JALA, a pretraining framework that learns Jointly-Aligned Latent Actions. JALA bypasses full visual dynamic reconstruction, instead learns a predictive action embedding aligned with both inverse dynamics and real actions. This yields a transition-aware, behavior-centric latent space for learning from heterogeneous human data. We scale this approach with UniHand-Mix, a 7.5M video corpus (>2,000 hours) blending laboratory and in-the-wild footage. Experiments demonstrate that JALA generates more realistic hand motions in both controlled and unconstrained scenarios, significantly improving downstream robot manipulation performance in both simulation and real-world tasks. These results indicate that jointly-aligned latent actions offer a scalable pathway for VLA pretraining from human data.

cs.RO↗

arXiv2026

Robust Kaczmarz methods for nearly singular linear systems

The Kaczmarz method is an efficient iterative algorithm for large-scale linear systems. However, its linear convergence rate suffers from ill-conditioned problems and is highly sensitive to the smallest nonzero singular value. In this work, we aim to extend the classical Kaczmarz to nearly singular linear systems that are row rank-deficient. We introduce a new concept of nearly singular property by treating the row space as an unstable subspace in the Grassman manifold. We then define a related important space called the approximate kernel, based on which a robust kernel-augmented Kaczmarz (KaK) is introduced via the subspace correction framework and analyzed by the well-known Xu--Zikatanov identity. To get an implementable version, we further introduce the approximate dual kernel and transform KaK into an equivalent kernel-augmented coordinate descent. Furthermore, we develop an accelerated variant and establish the improved rate of convergence matching the optimal complexity of first-order methods. Compared with existing methods, ours achieve uniform convergence rates for nearly singular linear systems, and the robustness has been confirmed by some numerical tests.

math.NA↗

arXiv2026

Rethinking Visual-Language-Action Model Scaling: Alignment, Mixture, and Regularization

While Vision-Language-Action (VLA) models show strong promise for generalist robot control, it remains unclear whether -- and under what conditions -- the standard "scale data" recipe translates to robotics, where training data is inherently heterogeneous across embodiments, sensors, and action spaces. We present a systematic, controlled study of VLA scaling that revisits core training choices for pretraining across diverse robots. Using a representative VLA framework that combines a vision-language backbone with flow-matching, we ablate key design decisions under matched conditions and evaluate in extensive simulation and real-robot experiments. To improve the reliability of real-world results, we introduce a Grouped Blind Ensemble protocol that blinds operators to model identity and separates policy execution from outcome judgment, reducing experimenter bias. Our analysis targets three dimensions of VLA scaling. (1) Physical alignment: we show that a unified end-effector (EEF)-relative action representation is critical for robust cross-embodiment transfer. (2) Embodiment mixture: we find that naively pooling heterogeneous robot datasets often induces negative transfer rather than gains, underscoring the fragility of indiscriminate data scaling. (3) Training regularization: we observe that intuitive strategies, such as sensory dropout and multi-stage fine-tuning, do not consistently improve performance at scale. Together, this study challenge some common assumptions about embodied scaling and provide practical guidance for training large-scale VLA policies from diverse robotic data. Project website: https://research.beingbeyond.com/rethink_vla

cs.RO↗

arXiv2026

Accelerated Mirror Descent Method through Variable and Operator Splitting

Mirror descent uses the mirror function to encode geometry and constraints, improving convergence while preserving feasibility. Accelerated Mirror Descent Methods (Acc-MD) are derived from a discretization of an accelerated mirror ODE system using a variable--operator splitting framework. A geometric assumption, termed the Generalized Cauchy-Schwarz (GCS) condition, is introduced to quantify the compatibility between the objective and the mirror geometry, under which the first accelerated linear convergence for Acc-MD on a broad class of problems is established. Numerical experiments on smooth and composite optimization tasks demonstrate that Acc-MD consistently outperforms existing accelerated variants, both theoretically and empirically.

math.OC↗

arXiv2026

Shear-Induced Wobbling and Motility Suppression in Swimming Bacteria

The intricate wobbling motion of flagellated bacteria, characterized by the periodic precession of the cell body, is a determinant factor in their motility and navigation within complex fluid environments. While well-studied in quiescent fluids, bacterial wobbling under ubiquitous flow conditions remains unexplored. In this work, we investigate the wobbling dynamics of \textit{Escherichia coli} swimming near surfaces under steady shear flow. Our experiments reveal that the wobbling amplitude intensifies with flow strength before reaching a plateau, with this amplification exhibiting a strong dependence on the swimming orientation relative to the flow direction. It turns out that the enhanced wobbling remains governed by the misalignment between the cell body and the flagellar bundle. Furthermore, we observe that the wobbling frequency increases monotonically with flow strength, and that shorter bacteria exhibit more pronounced variations in both amplitude and frequency. By linking the wobbling motion to the intrinsic body-flagella misalignment, we attribute the flow-enhanced precession to a combination of shear- and chirality-induced torques acting on the flexible flagellar hook. This mechanical coupling ultimately suppresses the net migration velocity as the flow rate increases. These findings elucidate the elastohydrodynamic mechanisms by which shear flow modifies bacterial locomotion near surfaces, with implications for microbial transport in physiological and ecological environments.

cond-mat.soft↗

arXiv2026

Being-H0.5: Scaling Human-Centric Robot Learning for Cross-Embodiment Generalization

We introduce Being-H0.5, a foundational Vision-Language-Action (VLA) model designed for robust cross-embodiment generalization across diverse robotic platforms. While existing VLAs often struggle with morphological heterogeneity and data scarcity, we propose a human-centric learning paradigm that treats human interaction traces as a universal "mother tongue" for physical interaction. To support this, we present UniHand-2.0, the largest embodied pre-training recipe to date, comprising over 35,000 hours of multimodal data across 30 distinct robotic embodiments. Our approach introduces a Unified Action Space that maps heterogeneous robot controls into semantically aligned slots, enabling low-resource robots to bootstrap skills from human data and high-resource platforms. Built upon this human-centric foundation, we design a unified sequential modeling and multi-task pre-training paradigm to bridge human demonstrations and robotic execution. Architecturally, Being-H0.5 utilizes a Mixture-of-Transformers design featuring a novel Mixture-of-Flow (MoF) framework to decouple shared motor primitives from specialized embodiment-specific experts. Finally, to make cross-embodiment policies stable in the real world, we introduce Manifold-Preserving Gating for robustness under sensory shift and Universal Async Chunking to universalize chunked control across embodiments with different latency and control profiles. We empirically demonstrate that Being-H0.5 achieves state-of-the-art results on simulated benchmarks, such as LIBERO (98.9%) and RoboCasa (53.9%), while also exhibiting strong cross-embodiment capabilities on five robotic platforms.

cs.RO↗

arXiv2026

Mistake Notebook Learning: Batch-Clustered Failures for Training-Free Agent Adaptation

With the growing adoption of Large Language Model (LLM) agents in persistent, real-world roles, they naturally encounter continuous streams of tasks and inevitable failures. A key limitation, however, is their inability to systematically learn from these mistakes, forcing them to repeat identical errors in similar contexts. Unlike prior training-free methods that primarily store raw instance-level experience or focus on retrieving successful trajectories, we propose Mistake Notebook Learning (MNL), a novel memory framework that enables agents to self-curate generalizable guidance from batch-clustered failures. This mechanism allows agents to distill shared error patterns into structured "mistake notes," updating an external memory only when batch performance improves to ensure stability. To further amplify adaptability, we integrate MNL with test-time scaling, leveraging aggregated failure patterns to actively steer the search process away from known pitfalls. Experiments on mathematical reasoning, Text-to-SQL, and interactive agent benchmarks show that MNL achieves competitive performance compared to existing memory mechanisms and in-context methods in both effectiveness and efficiency. These findings position structured mistake abstraction as a critical lever for robust agent evolution, enabling continuous improvement without the cost of parameter updates. The code is available at https://github.com/Bairong-Xdynamics/MistakeNotebookLearning/tree/main.

cs.CL↗

arXiv2026

DyDiT++: Diffusion Transformers with Timestep and Spatial Dynamics for Efficient Visual Generation

Diffusion Transformer (DiT), an emerging diffusion model for visual generation, has demonstrated superior performance but suffers from substantial computational costs. Our investigations reveal that these costs primarily stem from the static inference paradigm, which inevitably introduces redundant computation in certain diffusion timesteps and spatial regions. To overcome this inefficiency, we propose Dynamic Diffusion Transformer (DyDiT), an architecture that dynamically adjusts its computation along both timestep and spatial dimensions. Building on these designs, we present an extended version, DyDiT++, with improvements in three key aspects. First, it extends the generation mechanism of DyDiT beyond diffusion to flow matching, demonstrating that our method can also accelerate flow-matching-based generation, enhancing its versatility. Furthermore, we enhance DyDiT to tackle more complex visual generation tasks, including video generation and text-to-image generation, thereby broadening its real-world applications. Finally, to address the high cost of full fine-tuning and democratize technology access, we investigate the feasibility of training DyDiT in a parameter-efficient manner and introduce timestep-based dynamic LoRA (TD-LoRA). Extensive experiments on diverse visual generation models, including DiT, SiT, Latte, and FLUX, demonstrate the effectiveness of DyDiT++. Remarkably, with <3% additional fine-tuning iterations, our approach reduces the FLOPs of DiT-XL by 51%, yielding 1.73x realistic speedup on hardware, and achieves a competitive FID score of 2.07 on ImageNet. The code is available at https://github.com/alibaba-damo-academy/DyDiT.

cs.CV↗

arXiv2025

Spatial-Aware VLA Pretraining through Visual-Physical Alignment from Human Videos

Vision-Language-Action (VLA) models provide a promising paradigm for robot learning by integrating visual perception with language-guided policy learning. However, most existing approaches rely on 2D visual inputs to perform actions in 3D physical environments, creating a significant gap between perception and action grounding. To bridge this gap, we propose a Spatial-Aware VLA Pretraining paradigm that performs explicit alignment between visual space and physical space during pretraining, enabling models to acquire 3D spatial understanding before robot policy learning. Starting from pretrained vision-language models, we leverage large-scale human demonstration videos to extract 3D visual and 3D action annotations, forming a new source of supervision that aligns 2D visual observations with 3D spatial reasoning. We instantiate this paradigm with VIPA-VLA, a dual-encoder architecture that incorporates a 3D visual encoder to augment semantic visual representations with 3D-aware features. When adapted to downstream robot tasks, VIPA-VLA achieves significantly improved grounding between 2D vision and 3D action, resulting in more robust and generalizable robotic policies.

cs.RO↗

arXiv2025

PlayerOne: Egocentric World Simulator

We introduce PlayerOne, the first egocentric realistic world simulator, facilitating immersive and unrestricted exploration within vividly dynamic environments. Given an egocentric scene image from the user, PlayerOne can accurately construct the corresponding world and generate egocentric videos that are strictly aligned with the real scene human motion of the user captured by an exocentric camera. PlayerOne is trained in a coarse-to-fine pipeline that first performs pretraining on large-scale egocentric text-video pairs for coarse-level egocentric understanding, followed by finetuning on synchronous motion-video data extracted from egocentric-exocentric video datasets with our automatic construction pipeline. Besides, considering the varying importance of different components, we design a part-disentangled motion injection scheme, enabling precise control of part-level movements. In addition, we devise a joint reconstruction framework that progressively models both the 4D scene and video frames, ensuring scene consistency in the long-form video generation. Experimental results demonstrate its great generalization ability in precise control of varying human movements and worldconsistent modeling of diverse scenarios. It marks the first endeavor into egocentric real-world simulation and can pave the way for the community to delve into fresh frontiers of world modeling and its diverse applications.

cs.CV↗

arXiv2025

Digital Twin Aided Millimeter Wave MIMO: Site-Specific Beam Codebook Learning

Learning site-specific beams that adapt to the deployment environment, interference sources, and hardware imperfections can lead to noticeable performance gains in coverage, data rate, and power saving, among other interesting advantages. This learning process, however, typically requires a large number of active interactions/iterations, which limits its practical feasibility and leads to excessive overhead. To address these challenges, we propose a digital twin aided codebook learning framework, where a site-specific digital twin is leveraged to generate synthetic channel data for codebook learning. We also propose to learn separate codebooks for line-of-sight and non-line-of-sight users, leveraging the geometric information provided by the digital twin. Simulation results demonstrate that the codebook learned from the digital twin can adapt to the environment geometry and user distribution, leading to high received signal-to-noise ratio performance. Moreover, we identify the ray-tracing accuracy as the most critical factor in digital twin fidelity that impacts the learned codebook performance.

eess.SP↗

arXiv2025

Lumine: An Open Recipe for Building Generalist Agents in 3D Open Worlds

We introduce Lumine, the first open recipe for developing generalist agents capable of completing hours-long complex missions in real time within challenging 3D open-world environments. Lumine adopts a human-like interaction paradigm that unifies perception, reasoning, and action in an end-to-end manner, powered by a vision-language model. It processes raw pixels at 5 Hz to produce precise 30 Hz keyboard-mouse actions and adaptively invokes reasoning only when necessary. Trained in Genshin Impact, Lumine successfully completes the entire five-hour Mondstadt main storyline on par with human-level efficiency and follows natural language instructions to perform a broad spectrum of tasks in both 3D open-world exploration and 2D GUI manipulation across collection, combat, puzzle-solving, and NPC interaction. In addition to its in-domain performance, Lumine demonstrates strong zero-shot cross-game generalization. Without any fine-tuning, it accomplishes 100-minute missions in Wuthering Waves and the full five-hour first chapter of Honkai: Star Rail. These promising results highlight Lumine's effectiveness across distinct worlds and interaction dynamics, marking a concrete step toward generalist agents in open-ended environments.

cs.AI↗

arXiv2025

Generative Decoding of Compressed CSI for MIMO Precoding Design

Massive MIMO systems can enhance spectral and energy efficiency, but they require accurate channel state information (CSI), which becomes costly as the number of antennas increases. While machine learning (ML) autoencoders show promise for CSI reconstruction and reducing feedback overhead, they introduce new challenges with standardization, interoperability, and backward compatibility. Also, the significant data collection needed for training makes real-world deployment difficult. To overcome these drawbacks, we propose an ML-based, decoder-only solution for compressed CSI. Our approach uses a standardized encoder for CSI compression on the user side and a site-specific generative decoder at the base station to refine the compressed CSI using environmental knowledge. We introduce two training schemes for the generative decoder: An end-to-end method and a two-stage method, both utilizing a goal-oriented loss function. Furthermore, we reduce the data collection overhead by using a site-specific digital twin to generate synthetic CSI data for training. Our simulations highlight the effectiveness of this solution across various feedback overhead regimes.

eess.SP↗

arXiv2025

An accelerated primal-dual flow for linearly constrained multiobjective optimization

In this paper, we propose a continuous-time primal-dual approach for linearly constrained multiobjective optimization problems. A novel dynamical model, called accelerated multiobjective primal-dual flow, is presented with a second-order equation for the primal variable and a first-order equation for the dual variable. It can be viewed as an extension of the accelerated primal-dual flow by Luo [arXiv:2109.12604, 2021] for the single objective case. To facilitate the convergence rate analysis, we introduce a new merit function, which motivates the use of the feasibility violation and the objective gap to measure the weakly Pareto optimality. By using a proper Lyapunov function, we establish the exponential decay rate in the continuous level. After that, we consider an implicit-explicit scheme, which yields an accelerated multiobjective primal-dual method with a quadratic subproblem, and prove the sublinear rates of the feasibility violation and the objective gap, under the convex case and the strongly convex case, respectively. Numerical results are provided to demonstrate the performance of the proposed method.

math.OC↗

arXiv2025

Digital Twin Aided Massive MIMO CSI Feedback: Exploring the Impact of Twinning Fidelity

Deep learning (DL) techniques have demonstrated strong performance in compressing and reconstructing channel state information (CSI) while reducing feedback overhead in massive MIMO systems. A key challenge, however, is their reliance on extensive site-specific training data, whose real-world collection incurs significant overhead and limits scalability across deployment sites. To address this, we propose leveraging site-specific digital twins to assist the training of DL-based CSI compression models. The digital twin integrates an electromagnetic (EM) 3D model of the environment, a hardware model, and ray tracing to produce site-specific synthetic CSI data, allowing DL models to be trained without the need for extensive real-world measurements. We further develop a fidelity analysis framework that decomposes digital twin quality into four key aspects: 3D geometry, material properties, ray tracing, and hardware modeling. We explore how these factors influence the reliability of the data and model performance. To enhance the adaptability to real-world environments, we propose a refinement strategy that incorporates a limited amount of real-world data to fine-tune the DL model pre-trained on the digital twin dataset. Evaluation results show that models trained on site-specific digital twins outperform those trained on generic datasets, with the proposed refinement method effectively enhancing performance using limited real-world data. The simulations also highlight the importance of digital twin fidelity, especially in 3D geometry, ray tracing, and hardware modeling, for improving CSI reconstruction quality. This analysis framework offers valuable insights into the critical fidelity aspects, and facilitates more efficient digital twin development and deployment strategies for various wireless communication tasks.

eess.SP↗

arXiv2025

Light-SQ: Structure-aware Shape Abstraction with Superquadrics for Generated Meshes

In user-generated-content (UGC) applications, non-expert users often rely on image-to-3D generative models to create 3D assets. In this context, primitive-based shape abstraction offers a promising solution for UGC scenarios by compressing high-resolution meshes into compact, editable representations. Towards this end, effective shape abstraction must therefore be structure-aware, characterized by low overlap between primitives, part-aware alignment, and primitive compactness. We present Light-SQ, a novel superquadric-based optimization framework that explicitly emphasizes structure-awareness from three aspects. (a) We introduce SDF carving to iteratively udpate the target signed distance field, discouraging overlap between primitives. (b) We propose a block-regrow-fill strategy guided by structure-aware volumetric decomposition, enabling structural partitioning to drive primitive placement. (c) We implement adaptive residual pruning based on SDF update history to surpress over-segmentation and ensure compact results. In addition, Light-SQ supports multiscale fitting, enabling localized refinement to preserve fine geometric details. To evaluate our method, we introduce 3DGen-Prim, a benchmark extending 3DGen-Bench with new metrics for both reconstruction quality and primitive-level editability. Extensive experiments demonstrate that Light-SQ enables efficient, high-fidelity, and editable shape abstraction with superquadrics for complex generated geometry, advancing the feasibility of 3D UGC creation.

cs.GR↗

arXiv2025

CoreMark: Toward Robust and Universal Text Watermarking Technique

Text watermarking schemes have gained considerable attention in recent years, yet still face critical challenges in achieving simultaneous robustness, generalizability, and imperceptibility. This paper introduces a new embedding paradigm,termed CORE, which comprises several consecutively aligned black pixel segments. Its key innovation lies in its inherent noise resistance during transmission and broad applicability across languages and fonts. Based on the CORE, we present a text watermarking framework named CoreMark. Specifically, CoreMark first dynamically extracts COREs from characters. Then, the characters with stronger robustness are selected according to the lengths of COREs. By modifying the thickness of the CORE, the hidden data is embedded into the selected characters without causing significant visual distortions. Moreover, a general plug-and-play embedding strength modulator is proposed, which can adaptively enhance the robustness for small font sizes by adjusting the embedding strength according to the font size. Experimental evaluation indicates that CoreMark demonstrates outstanding generalizability across multiple languages and fonts. Compared to existing methods, CoreMark achieves significant improvements in resisting screenshot, print-scan, and print camera attacks, while maintaining satisfactory imperceptibility.

cs.CV↗

arXiv2025

Preacher: Paper-to-Video Agentic System

The paper-to-video task converts a research paper into a structured video abstract, distilling key concepts, methods, and conclusions into an accessible, well-organized format. While state-of-the-art video generation models demonstrate potential, they are constrained by limited context windows, rigid video duration constraints, limited stylistic diversity, and an inability to represent domain-specific knowledge. To address these limitations, we introduce Preacher, the first paper-to-video agentic system. Preacher employs a topdown approach to decompose, summarize, and reformulate the paper, followed by bottom-up video generation, synthesizing diverse video segments into a coherent abstract. To align cross-modal representations, we define key scenes and introduce a Progressive Chain of Thought (P-CoT) for granular, iterative planning. Preacher successfully generates high-quality video abstracts across five research fields, demonstrating expertise beyond current video generation models. Code will be released at: https://github.com/Gen-Verse/Paper2Video

cs.CV↗