arXiv subjects

Hongtao Wu

Publications and source records attributed to Hongtao Wu.

At least 19 recordsLinked to original sources

arXiv2026

BridgeVLA++: A Data-Efficient, Generalizable, and Memory-Augmented Vision-Language-Action Framework for 3D Manipulation

Leveraging pre-trained vision-language models (VLMs) to construct vision-language-action (VLA) models has emerged as a promising paradigm for 3D robot manipulation. However, existing 3D VLA methods remain data-hungry, exhibit limited generalization under distribution shifts, and lack explicit memory of past observations. These limitations hinder their application to data-scarce, open-world, and memory-dependent manipulation scenarios. Our previous work, BridgeVLA, improves data efficiency and generalization by preserving the input--output alignment of a pre-trained VLM during 3D action learning: raw point clouds are projected into multi-view images, and intermediate heatmaps are predicted before generating robot actions. In this work, we develop BridgeVLA++ by equipping BridgeVLA with a unified spatio-temporal memory architecture that models persistent spatial context and temporal interaction history. The resulting memory-augmented framework can reason over observation histories while preserving BridgeVLA's data efficiency and generalization capabilities. Extensive experiments show that our framework achieves strong performance on spatial manipulation tasks while exhibiting robust generalization. BridgeVLA++ further achieves state-of-the-art performance on two challenging memory-dependent manipulation benchmarks without sacrificing the data efficiency and generalization of the original BridgeVLA. In addition, BridgeVLA++ performs effectively in bimanual manipulation settings and is validated on an additional real-world robotic platform, demonstrating its scalability across tasks, environments, and robotic platforms. These results establish BridgeVLA++ as a unified 3D vision-language-action framework that simultaneously supports data-efficient learning, robust generalization, and effective memory-aware robot manipulation. Project website: https://bridgevla-plus.github.io/.

cs.RO↗

arXiv2026

Towards Autonomous and Auditable Medical Imaging Model Development

Large language model (LLM) agents are beginning to automate machine learning engineering (MLE) by coupling planning, code execution, debugging, and empirical feedback. Translating this capability to medical imaging remains difficult because each task imposes modality-specific experimentation and strict requirements for validation protocols and prediction artifacts. Here we introduce AMID, an autonomous multi-agent framework for medical imaging model development. AMID first proposes Data-Conditioned Method Planning, which refines coarse task-level search spaces into executable, parallelizable method lanes grounded in task-specific data analysis and runnable medical-imaging resources. It then develops Verification-Guided Two-Stage Optimization, moving from broad early exploration of diverse method lanes to selective exploitation of promising candidates while enforcing strict verification of validation protocols, metric computation, and prediction artifacts throughout the optimization. Across 20 medical imaging challenge tasks spanning diverse modalities and prediction types, AMID outperformed evaluated general-purpose MLE systems and, on several tasks, approached or matched strong human-designed challenge solutions. These results suggest that AMID can turn task-specific medical imaging model development from bespoke manual engineering into an agentic workflow for producing high-performing and auditable model artifacts across heterogeneous tasks.

cs.CV↗

arXiv2026

Towards a Medical AI Scientist

Autonomous systems that generate scientific hypotheses, conduct experiments, and draft manuscripts have recently emerged as a promising paradigm for accelerating discovery. However, existing AI Scientists remain largely domain-agnostic, limiting their applicability to clinical medicine, where research is required to be grounded in medical evidence with specialized data modalities. In this work, we introduce Medical AI Scientist, the first autonomous research framework tailored to clinical autonomous research. It enables clinically grounded ideation by transforming extensively surveyed literature into actionable evidence through clinician-engineer co-reasoning mechanism, which improves the traceability of generated research ideas. It further facilitates evidence-grounded manuscript drafting guided by structured medical compositional conventions and ethical policies. The framework operates under 3 research modes, namely paper-based reproduction, literature-inspired innovation, and task-driven exploration, each corresponding to a distinct level of automated scientific inquiry with progressively increasing autonomy. Comprehensive evaluations by both large language models and human experts demonstrate that the ideas generated by the Medical AI Scientist are of substantially higher quality than those produced by commercial LLMs across 171 cases, 19 clinical tasks, and 6 data modalities. Meanwhile, our system achieves strong alignment between the proposed method and its implementation, while also demonstrating significantly higher success rates in executable experiments. Double-blind evaluations by human experts and the Stanford Agentic Reviewer suggest that the generated manuscripts approach MICCAI-level quality, while consistently surpassing those from ISBI and BIBM. The proposed Medical AI Scientist highlights the potential of leveraging AI for autonomous scientific discovery in healthcare.

cs.AI↗

arXiv2025

Human-assisted Robotic Policy Refinement via Action Preference Optimization

Establishing a reliable and iteratively refined robotic system is essential for deploying real-world applications. While Vision-Language-Action (VLA) models are widely recognized as the foundation model for such robotic deployment, their reliance on offline expert demonstrations critically limits their capacity for post-deployment refinement. To mitigate this limitation, we introduce Action Preference Optimization (APO), a method designed to refine VLA models by human-assisted preference alignment gathered through interaction with environments. This method begins with a human-robot collaboration framework for reliable failure correction and interaction trajectory collection through human intervention. However, directly leveraging these interaction trajectories for preference optimization is non-trivial due to the challenges of irreversible robotic actions and token distribution mismatch. To solve this, APO proposes an adaptive reweighting algorithm with binary desirability signals derived from interaction, empowering VLA models effectively suppress failure-prone actions while enhancing corrective action adaptation. Ultimately, APO equips VLA models with the crucial capability to learn from failure, paving the way for their iterative refinement and reliable deployment in dynamic environments. The experiments conducted in simulation and real-world scenarios prove superior generalization and robustness of our human-assisted framework across a variety of manipulation tasks. We believe this work could bring insights for efficient and stable optimization of VLA models through human-robot collaboration. The code and dataset are released at https://github.com/GeWu-Lab/Action-Preference-Optimization

cs.RO↗

arXiv2025

BridgeVLA: Input-Output Alignment for Efficient 3D Manipulation Learning with Vision-Language Models

Recently, leveraging pre-trained vision-language models (VLMs) for building vision-language-action (VLA) models has emerged as a promising approach to effective robot manipulation learning. However, only few methods incorporate 3D signals into VLMs for action prediction, and they do not fully leverage the spatial structure inherent in 3D data, leading to low sample efficiency. In this paper, we introduce BridgeVLA, a novel 3D VLA model that (1) projects 3D inputs to multiple 2D images, ensuring input alignment with the VLM backbone, and (2) utilizes 2D heatmaps for action prediction, unifying the input and output spaces within a consistent 2D image space. In addition, we propose a scalable pre-training method that equips the VLM backbone with the capability to predict 2D heatmaps before downstream policy learning. Extensive experiments show the proposed method is able to learn 3D manipulation efficiently and effectively. BridgeVLA outperforms state-of-the-art baseline methods across three simulation benchmarks. In RLBench, it improves the average success rate from 81.4% to 88.2%. In COLOSSEUM, it demonstrates significantly better performance in challenging generalization settings, boosting the average success rate from 56.7% to 64.0%. In GemBench, it surpasses all the comparing baseline methods in terms of average success rate. In real-robot experiments, BridgeVLA outperforms a state-of-the-art baseline method by 32% on average. It generalizes robustly in multiple out-of-distribution settings, including visual disturbances and unseen instructions. Remarkably, it is able to achieve a success rate of 96.8% on 10+ tasks with only 3 trajectories per task, highlighting its extraordinary sample efficiency. Project Website:https://bridgevla.github.io/

cs.RO↗

arXiv2025

Beyond Tag Collision: Cluster-based Memory Management for Tag-based Sanitizers

Tag-based sanitizers attach a small "key" to each pointer and a matching "lock" tag to its target memory object, enabling runtime verification of pointer-object consistency and helping developers to detect potential memory violations. However, the limited tag encoding space challenges existing studies in assigning distinct tags to memory objects across temporal and spatial dimensions, leading to potential tag collisions. In this paper, we present ClusterTag, a novel cluster-based memory allocator aimed at simultaneously mitigating tag collisions in both temporal and spatial dimensions. The core design of ClusterTag effectively balances the significant mismatch between tag encoding space and memory objects: it divides memory objects into multiple independent clusters, thereby limiting tag collisions to finite chunks within each cluster. To mitigate tag collisions across clusters, we design a cluster-grained heap randomization scheme. This approach introduces random address intervals between clusters and further breaks the entropy limitation of the tag space. ClusterTag has been implemented as an independent memory allocator that seamlessly integrates with tag-based sanitizers such as HWASan, and maintains comparable performance overhead (within 1%) at various randomization densities. Security evaluations on the Juliet dataset indicate that ClusterTag exhibits deterministic results across 500 repeated tests (5,652 reported and 1,530 missed), while the existing three types of tag assignment strategies all exhibit probabilistic false negatives due to tag collisions. Quantitative analysis across three tag collision distance metrics-minimum, average, and unpredictability-demonstrates that ClusterTag achieves balanced improvements across all three, whereas prior tag assignment schemes (random, staggered, fixed) show significant trade-offs in at least one metric.

cs.CR↗

arXiv2025

Single Domain Generalization for Multimodal Cross-Cancer Prognosis via Dirac Rebalancer and Distribution Entanglement

Deep learning has shown remarkable performance in integrating multimodal data for survival prediction. However, existing multimodal methods mainly focus on single cancer types and overlook the challenge of generalization across cancers. In this work, we are the first to reveal that multimodal prognosis models often generalize worse than unimodal ones in cross-cancer scenarios, despite the critical need for such robustness in clinical practice. To address this, we propose a new task: Cross-Cancer Single Domain Generalization for Multimodal Prognosis, which evaluates whether models trained on a single cancer type can generalize to unseen cancers. We identify two key challenges: degraded features from weaker modalities and ineffective multimodal integration. To tackle these, we introduce two plug-and-play modules: Sparse Dirac Information Rebalancer (SDIR) and Cancer-aware Distribution Entanglement (CADE). SDIR mitigates the dominance of strong features by applying Bernoulli-based sparsification and Dirac-inspired stabilization to enhance weaker modality signals. CADE, designed to synthesize the target domain distribution, fuses local morphological cues and global gene expression in latent space. Experiments on a four-cancer-type benchmark demonstrate superior generalization, laying the foundation for practical, robust cross-cancer multimodal prognosis. Code is available at https://github.com/HopkinsKwong/MCCSDG

cs.CV↗

arXiv2025

BrainSegDMlF: A Dynamic Fusion-enhanced SAM for Brain Lesion Segmentation

The segmentation of substantial brain lesions is a significant and challenging task in the field of medical image segmentation. Substantial brain lesions in brain imaging exhibit high heterogeneity, with indistinct boundaries between lesion regions and normal brain tissue. Small lesions in single slices are difficult to identify, making the accurate and reproducible segmentation of abnormal regions, as well as their feature description, highly complex. Existing methods have the following limitations: 1) They rely solely on single-modal information for learning, neglecting the multi-modal information commonly used in diagnosis. This hampers the ability to comprehensively acquire brain lesion information from multiple perspectives and prevents the effective integration and utilization of multi-modal data inputs, thereby limiting a holistic understanding of lesions. 2) They are constrained by the amount of data available, leading to low sensitivity to small lesions and difficulty in detecting subtle pathological changes. 3) Current SAM-based models rely on external prompts, which cannot achieve automatic segmentation and, to some extent, affect diagnostic efficiency.To address these issues, we have developed a large-scale fully automated segmentation model specifically designed for brain lesion segmentation, named BrainSegDMLF. This model has the following features: 1) Dynamic Modal Interactive Fusion (DMIF) module that processes and integrates multi-modal data during the encoding process, providing the SAM encoder with more comprehensive modal information. 2) Layer-by-Layer Upsampling Decoder, enabling the model to extract rich low-level and high-level features even with limited data, thereby detecting the presence of small lesions. 3) Automatic segmentation masks, allowing the model to generate lesion masks automatically without requiring manual prompts.

cs.CV↗

arXiv2025

CineVision: An Interactive Pre-visualization Storyboard System for Director-Cinematographer Collaboration

Effective communication between directors and cinematographers is fundamental in film production, yet traditional approaches relying on visual references and hand-drawn storyboards often lack the efficiency and precision necessary during pre-production. We present CineVision, an AI-driven platform that integrates scriptwriting with real-time visual pre-visualization to bridge this communication gap. By offering dynamic lighting control, style emulation based on renowned filmmakers, and customizable character design, CineVision enables directors to convey their creative vision with heightened clarity and rapidly iterate on scene composition. In a 24-participant lab study, CineVision yielded shorter task times and higher usability ratings than two baseline methods, suggesting a potential to ease early-stage communication and accelerate storyboard drafts under controlled conditions. These findings underscore CineVision's potential to streamline pre-production processes and foster deeper creative synergy among filmmaking teams, particularly for new collaborators. Our code and demo are available at https://github.com/TonyHongtaoWu/CineVision.

cs.HC↗

arXiv2025

IRASim: A Fine-Grained World Model for Robot Manipulation

World models allow autonomous agents to plan and explore by predicting the visual outcomes of different actions. However, for robot manipulation, it is challenging to accurately model the fine-grained robot-object interaction within the visual space using existing methods which overlooks precise alignment between each action and the corresponding frame. In this paper, we present IRASim, a novel world model capable of generating videos with fine-grained robot-object interaction details, conditioned on historical observations and robot action trajectories. We train a diffusion transformer and introduce a novel frame-level action-conditioning module within each transformer block to explicitly model and strengthen the action-frame alignment. Extensive experiments show that: (1) the quality of the videos generated by our method surpasses all the baseline methods and scales effectively with increased model size and computation; (2) policy evaluations using IRASim exhibit a strong correlation with those using the ground-truth simulator, highlighting its potential to accelerate real-world policy evaluation; (3) testing-time scaling through model-based planning with IRASim significantly enhances policy performance, as evidenced by an improvement in the IoU metric on the Push-T benchmark from 0.637 to 0.961; (4) IRASim provides flexible action controllability, allowing virtual robotic arms in datasets to be controlled via a keyboard or VR controller.

cs.RO↗

arXiv2025

GR-3 Technical Report

We report our recent progress towards building generalist robot policies, the development of GR-3. GR-3 is a large-scale vision-language-action (VLA) model. It showcases exceptional capabilities in generalizing to novel objects, environments, and instructions involving abstract concepts. Furthermore, it can be efficiently fine-tuned with minimal human trajectory data, enabling rapid and cost-effective adaptation to new settings. GR-3 also excels in handling long-horizon and dexterous tasks, including those requiring bi-manual manipulation and mobile movement, showcasing robust and reliable performance. These capabilities are achieved through a multi-faceted training recipe that includes co-training with web-scale vision-language data, efficient fine-tuning from human trajectory data collected via VR devices, and effective imitation learning with robot trajectory data. In addition, we introduce ByteMini, a versatile bi-manual mobile robot designed with exceptional flexibility and reliability, capable of accomplishing a wide range of tasks when integrated with GR-3. Through extensive real-world experiments, we show GR-3 surpasses the state-of-the-art baseline method, $π_0$, on a wide variety of challenging tasks. We hope GR-3 can serve as a step towards building generalist robots capable of assisting humans in daily life.

cs.RO↗

arXiv2025

NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images: Methods and Results

This paper reviews the NTIRE 2025 Challenge on Day and Night Raindrop Removal for Dual-Focused Images. This challenge received a wide range of impressive solutions, which are developed and evaluated using our collected real-world Raindrop Clarity dataset. Unlike existing deraining datasets, our Raindrop Clarity dataset is more diverse and challenging in degradation types and contents, which includes day raindrop-focused, day background-focused, night raindrop-focused, and night background-focused degradations. This dataset is divided into three subsets for competition: 14,139 images for training, 240 images for validation, and 731 images for testing. The primary objective of this challenge is to establish a new and powerful benchmark for the task of removing raindrops under varying lighting and focus conditions. There are a total of 361 participants in the competition, and 32 teams submitting valid solutions and fact sheets for the final testing phase. These submissions achieved state-of-the-art (SOTA) performance on the Raindrop Clarity dataset. The project can be found at https://lixinustc.github.io/CVPR-NTIRE2025-RainDrop-Competition.github.io/.

cs.CV↗

arXiv2025

MFM-DA: Instance-Aware Adaptor and Hierarchical Alignment for Efficient Domain Adaptation in Medical Foundation Models

Medical Foundation Models (MFMs), trained on large-scale datasets, have demonstrated superior performance across various tasks. However, these models still struggle with domain gaps in practical applications. Specifically, even after fine-tuning on source-domain data, task-adapted foundation models often perform poorly in the target domain. To address this challenge, we propose a few-shot unsupervised domain adaptation (UDA) framework for MFMs, named MFM-DA, which only leverages a limited number of unlabeled target-domain images. Our approach begins by training a Denoising Diffusion Probabilistic Model (DDPM), which is then adapted to the target domain using a proposed dynamic instance-aware adaptor and a distribution direction loss, enabling the DDPM to translate source-domain images into the target domain style. The adapted images are subsequently processed through the MFM, where we introduce a designed channel-spatial alignment Low-Rank Adaptation (LoRA) to ensure effective feature alignment. Extensive experiments on optic cup and disc segmentation tasks demonstrate that MFM-DA outperforms state-of-the-art methods. Our work provides a practical solution to the domain gap issue in real-world MFM deployment. Code will be available at here.

cs.CV↗

arXiv2024

GR-MG: Leveraging Partially Annotated Data via Multi-Modal Goal-Conditioned Policy

The robotics community has consistently aimed to achieve generalizable robot manipulation with flexible natural language instructions. One primary challenge is that obtaining robot trajectories fully annotated with both actions and texts is time-consuming and labor-intensive. However, partially-annotated data, such as human activity videos without action labels and robot trajectories without text labels, are much easier to collect. Can we leverage these data to enhance the generalization capabilities of robots? In this paper, we propose GR-MG, a novel method which supports conditioning on a text instruction and a goal image. During training, GR-MG samples goal images from trajectories and conditions on both the text and the goal image or solely on the image when text is not available. During inference, where only the text is provided, GR-MG generates the goal image via a diffusion-based image-editing model and conditions on both the text and the generated image. This approach enables GR-MG to leverage large amounts of partially-annotated data while still using languages to flexibly specify tasks. To generate accurate goal images, we propose a novel progress-guided goal image generation model which injects task progress information into the generation process. In simulation experiments, GR-MG improves the average number of tasks completed in a row of 5 from 3.35 to 4.04. In real-robot experiments, GR-MG is able to perform 58 different tasks and improves the success rate from 68.7\% to 78.1\% and 44.4\% to 60.6\% in simple and generalization settings, respectively. It also outperforms comparing baseline methods in few-shot learning of novel skills. Video demos, code, and checkpoints are available on the project page: https://gr-mg.github.io/.

cs.RO↗

arXiv2024

Semi-Supervised Video Desnowing Network via Temporal Decoupling Experts and Distribution-Driven Contrastive Regularization

Snow degradations present formidable challenges to the advancement of computer vision tasks by the undesirable corruption in outdoor scenarios. While current deep learning-based desnowing approaches achieve success on synthetic benchmark datasets, they struggle to restore out-of-distribution real-world snowy videos due to the deficiency of paired real-world training data. To address this bottleneck, we devise a new paradigm for video desnowing in a semi-supervised spirit to involve unlabeled real data for the generalizable snow removal. Specifically, we construct a real-world dataset with 85 snowy videos, and then present a Semi-supervised Video Desnowing Network (SemiVDN) equipped by a novel Distribution-driven Contrastive Regularization. The elaborated contrastive regularization mitigates the distribution gap between the synthetic and real data, and consequently maintains the desired snow-invariant background details. Furthermore, based on the atmospheric scattering model, we introduce a Prior-guided Temporal Decoupling Experts module to decompose the physical components that make up a snowy video in a frame-correlated manner. We evaluate our SemiVDN on benchmark datasets and the collected real snowy data. The experimental results demonstrate the superiority of our approach against state-of-the-art image- and video-level desnowing methods.

cs.CV↗

arXiv2024

GR-2: A Generative Video-Language-Action Model with Web-Scale Knowledge for Robot Manipulation

We present GR-2, a state-of-the-art generalist robot agent for versatile and generalizable robot manipulation. GR-2 is first pre-trained on a vast number of Internet videos to capture the dynamics of the world. This large-scale pre-training, involving 38 million video clips and over 50 billion tokens, equips GR-2 with the ability to generalize across a wide range of robotic tasks and environments during subsequent policy learning. Following this, GR-2 is fine-tuned for both video generation and action prediction using robot trajectories. It exhibits impressive multi-task learning capabilities, achieving an average success rate of 97.7% across more than 100 tasks. Moreover, GR-2 demonstrates exceptional generalization to new, previously unseen scenarios, including novel backgrounds, environments, objects, and tasks. Notably, GR-2 scales effectively with model size, underscoring its potential for continued growth and application. Project page: \url{https://gr2-manipulation.github.io}.

cs.RO↗

arXiv2024

World Model-based Perception for Visual Legged Locomotion

Legged locomotion over various terrains is challenging and requires precise perception of the robot and its surroundings from both proprioception and vision. However, learning directly from high-dimensional visual input is often data-inefficient and intricate. To address this issue, traditional methods attempt to learn a teacher policy with access to privileged information first and then learn a student policy to imitate the teacher's behavior with visual input. Despite some progress, this imitation framework prevents the student policy from achieving optimal performance due to the information gap between inputs. Furthermore, the learning process is unnatural since animals intuitively learn to traverse different terrains based on their understanding of the world without privileged knowledge. Inspired by this natural ability, we propose a simple yet effective method, World Model-based Perception (WMP), which builds a world model of the environment and learns a policy based on the world model. We illustrate that though completely trained in simulation, the world model can make accurate predictions of real-world trajectories, thus providing informative signals for the policy controller. Extensive simulated and real-world experiments demonstrate that WMP outperforms state-of-the-art baselines in traversability and robustness. Videos and Code are available at: https://wmp-loco.github.io/.

cs.RO↗

arXiv2024

RainMamba: Enhanced Locality Learning with State Space Models for Video Deraining

The outdoor vision systems are frequently contaminated by rain streaks and raindrops, which significantly degenerate the performance of visual tasks and multimedia applications. The nature of videos exhibits redundant temporal cues for rain removal with higher stability. Traditional video deraining methods heavily rely on optical flow estimation and kernel-based manners, which have a limited receptive field. Yet, transformer architectures, while enabling long-term dependencies, bring about a significant increase in computational complexity. Recently, the linear-complexity operator of the state space models (SSMs) has contrarily facilitated efficient long-term temporal modeling, which is crucial for rain streaks and raindrops removal in videos. Unexpectedly, its uni-dimensional sequential process on videos destroys the local correlations across the spatio-temporal dimension by distancing adjacent pixels. To address this, we present an improved SSMs-based video deraining network (RainMamba) with a novel Hilbert scanning mechanism to better capture sequence-level local information. We also introduce a difference-guided dynamic contrastive locality learning strategy to enhance the patch-level self-similarity learning ability of the proposed network. Extensive experiments on four synthesized video deraining datasets and real-world rainy videos demonstrate the effectiveness and efficiency of our network in the removal of rain streaks and raindrops. Our code and results are available at https://github.com/TonyHongtaoWu/RainMamba.

cs.CV↗