arXiv subjects

Tsun-Hsuan Wang

Publications and source records attributed to Tsun-Hsuan Wang.

At least 19 recordsLinked to original sources

arXiv2026

GS-Agent: Creating 4D Physical Worlds With Generative Simulation

Creating dynamic and physically realistic 4D worlds from natural language descriptions is both fascinating and challenging. Traditional computer graphics methods rely on manual creation, requiring extensive human effort to fine-tune materials, motions, and visual fidelity. Recent advances in generative foundation models have sparked interest in learning to generate such 4D worlds from large-scale data; however, existing methods still struggle to ensure physical plausibility and controllability. In this work, we take a different path by leveraging foundation models to construct an agentic system that emulates how humans traditionally create 4D worlds, yet automates the entire process. We present GS-Agent, an end-to-end multi-agent framework that integrates physics engines in the loop to generate realistic, dynamic, and controllable 4D physical worlds from natural language. Inspired by how humans build 4D worlds, GS-Agent decomposes the task into entity management, covering 3D asset curation, material tuning, placement, and motion control, and rendering configuration, including camera and lighting manipulation. Multiple agents with distinct expertise interact with the physics engine via code, seek multimodal feedback, and collaborate to iteratively construct 4D worlds that align with the given descriptions. Experimental results show that GS-Agent effectively converts natural language into diverse and physically plausible 4D worlds exhibiting rich interactions among liquids, deformable objects, and rigid bodies, while achieving cinematic camera and lighting control. We envision GS-Agent as a foundation for a new paradigm in 4D world generation, empowering creative content creation and physical AI. Project page at https://umass-embodied-agi.github.io/gs-agent/

cs.RO↗

arXiv2026

Action Images: End-to-End Policy Learning via Multiview Video Generation

World action models (WAMs) have emerged as a promising direction for robot policy learning, as they can leverage powerful video backbones to model the future states. However, existing approaches often rely on separate action modules, or use action representations that are not pixel-grounded, making it difficult to fully exploit the pretrained knowledge of video models and limiting transfer across viewpoints and environments. In this work, we present Action Images, a unified world action model that formulates policy learning as multiview video generation. Instead of encoding control as low-dimensional tokens, we translate 7-DoF robot actions into interpretable action images: multi-view action videos that are grounded in 2D pixels and explicitly track robot-arm motion. This pixel-grounded action representation allows the video backbone itself to act as a zero-shot policy, without a separate policy head or action module. Beyond control, the same unified model supports video-action joint generation, action-conditioned video generation, and action labeling under a shared representation. On RLBench and real-world evaluations, our model achieves the strongest zero-shot success rates and improves video-action joint generation quality over prior video-space world models, suggesting that interpretable action images are a promising route to policy learning.

cs.CV↗

arXiv2026

Human Insights Driven Latent Space for Different Driving Perspectives: A Unified Encoder for Efficient Multi-Task Inference

Autonomous driving systems require a comprehensive understanding of the environment, achieved by extracting visual features essential for perception, planning, and control. However, models trained solely on single-task objectives or generic datasets often lack the contextual information needed for robust performance in complex driving scenarios. In this work, we propose a unified encoder trained on multiple computer vision tasks crucial for urban driving, including depth, pose, and 3D scene flow estimation, as well as semantic, instance, panoptic, and motion segmentation. By integrating these diverse visual cues-similar to human perceptual mechanisms-the encoder captures rich features that enhance navigation-related predictions. We evaluate the model on steering estimation as a downstream task, leveraging its dense latent space. To ensure efficient multi-task learning, we introduce a multi-scale feature network for pose estimation and apply knowledge distillation from a multi-backbone teacher model. Our findings highlight two key findings: (1) the unified encoder achieves competitive performance across all visual perception tasks, demonstrating strong generalization capabilities; and (2) for steering estimation, the frozen unified encoder-leveraging dense latent representations-outperforms both its fine-tuned counterpart and the same frozen model pretrained on generic datasets like ImageNet. These results underline the significance of task-specific visual features and demonstrate the promise of multi-task learning in advancing autonomous driving systems. More details and the pretrained model are available at https://hi-computervision.github.io/uni-encoder/.

cs.CV↗

arXiv2026

PhyScensis: Physics-Augmented LLM Agents for Complex Physical Scene Arrangement

Automatically generating interactive 3D environments is crucial for scaling up robotic data collection in simulation. While prior work has primarily focused on 3D asset placement, it often overlooks the physical relationships between objects (e.g., contact, support, balance, and containment), which are essential for creating complex and realistic manipulation scenarios such as tabletop arrangements, shelf organization, or box packing. Compared to classical 3D layout generation, producing complex physical scenes introduces additional challenges: (a) higher object density and complexity (e.g., a small shelf may hold dozens of books), (b) richer supporting relationships and compact spatial layouts, and (c) the need to accurately model both spatial placement and physical properties. To address these challenges, we propose PhyScensis, an LLM agent-based framework powered by a physics engine, to produce physically plausible scene configurations with high complexity. Specifically, our framework consists of three main components: an LLM agent iteratively proposes assets with spatial and physical predicates; a solver, equipped with a physics engine, realizes these predicates into a 3D scene; and feedback from the solver informs the agent to refine and enrich the configuration. Moreover, our framework preserves strong controllability over fine-grained textual descriptions and numerical parameters (e.g., relative positions, scene stability), enabled through probabilistic programming for stability and a complementary heuristic that jointly regulates stability and spatial relations. Experimental results show that our method outperforms prior approaches in scene complexity, visual quality, and physical accuracy, offering a unified pipeline for generating complex physical scene layouts for robotic manipulation.

cs.RO↗

arXiv2026

See Less, Drive Better: Generalizable End-to-End Autonomous Driving via Foundation Models Stochastic Patch Selection

Recent advances in end-to-end autonomous driving show that policies trained on patch-aligned features extracted from foundation models generalize better to Out-of-Distribution (OOD). We hypothesize that due to the self-attention mechanism, each patch feature implicitly embeds/contains information from all other patches, represented in a different way and intensity, making these descriptors highly redundant. We quantify redundancy in such (BLIP2) features via PCA and cross-patch similarity: $90$% of variance is captured by $17/64$ principal components, and strong inter-token correlations are pervasive. Training on such overlapping information leads the policy to overfit spurious correlations, hurting OOD robustness. We present Stochastic-Patch-Selection (SPS), a simple yet effective approach for learning policies that are more robust, generalizable, and efficient. For every frame, SPS randomly masks a fraction of patch descriptors, not feeding them to the policy model, while preserving the spatial layout of the remaining patches. Thus, the policy is provided with different stochastic but complete views of the (same) scene: every random subset of patches acts like a different, yet still sensible, coherent projection of the world. The policy thus bases its decisions on features that are invariant to which specific tokens survive. Extensive experiments confirm that across all OOD scenarios, our method outperforms the state of the art (SOTA), achieving a $6.2$% average improvement and up to $20.4$% in closed-loop simulations, while being $2.4\times$ faster. We conduct ablations over masking rates and patch-feature reorganization, training and evaluating 9 systems, with 8 of them surpassing prior SOTA. Finally, we show that the same learned policy transfers to a physical, real-world car without any tuning.

cs.CV↗

arXiv2025

ReGen: Generative Robot Simulation via Inverse Design

Simulation plays a key role in scaling robot learning and validating policies, but constructing simulations remains a labor-intensive process. This paper introduces ReGen, a generative simulation framework that automates simulation design via inverse design. Given a robot's behavior -- such as a motion trajectory or an objective function -- and its textual description, ReGen infers plausible scenarios and environments that could have caused the behavior. ReGen leverages large language models to synthesize scenarios by expanding a directed graph that encodes cause-and-effect relationships, relevant entities, and their properties. This structured graph is then translated into a symbolic program, which configures and executes a robot simulation environment. Our framework supports (i) augmenting simulations based on ego-agent behaviors, (ii) controllable, counterfactual scenario generation, (iii) reasoning about agent cognition and mental states, and (iv) reasoning with distinct sensing modalities, such as braking due to faulty GPS signals. We demonstrate ReGen in autonomous driving and robot manipulation tasks, generating more diverse, complex simulated environments compared to existing simulations with high success rates, and enabling controllable generation for corner cases. This approach enhances the validation of robot policies and supports data or simulation augmentation, advancing scalable robot learning for improved generalization and robustness. We provide code and example videos at: https://regen-sim.github.io/

cs.RO↗

arXiv2025

Newton to Einstein: Axiom-Based Discovery via Game Design

This position paper argues that machine learning for scientific discovery should shift from inductive pattern recognition to axiom-based reasoning. We propose a game design framework in which scientific inquiry is recast as a rule-evolving system: agents operate within environments governed by axioms and modify them to explain outlier observations. Unlike conventional ML approaches that operate within fixed assumptions, our method enables the discovery of new theoretical structures through systematic rule adaptation. We demonstrate the feasibility of this approach through preliminary experiments in logic-based games, showing that agents can evolve axioms that solve previously unsolvable problems. This framework offers a foundation for building machine learning systems capable of creative, interpretable, and theory-driven discovery.

cs.CE↗

arXiv2025

Holistic Surgical Phase Recognition with Hierarchical Input Dependent State Space Models

Surgical workflow analysis is essential in robot-assisted surgeries, yet the long duration of such procedures poses significant challenges for comprehensive video analysis. Recent approaches have predominantly relied on transformer models; however, their quadratic attention mechanism restricts efficient processing of lengthy surgical videos. In this paper, we propose a novel hierarchical input-dependent state space model that leverages the linear scaling property of state space models to enable decision making on full-length videos while capturing both local and global dynamics. Our framework incorporates a temporally consistent visual feature extractor, which appends a state space model head to a visual feature extractor to propagate temporal information. The proposed model consists of two key modules: a local-aggregation state space model block that effectively captures intricate local dynamics, and a global-relation state space model block that models temporal dependencies across the entire video. The model is trained using a hybrid discrete-continuous supervision strategy, where both signals of discrete phase labels and continuous phase progresses are propagated through the network. Experiments have shown that our method outperforms the current state-of-the-art methods by a large margin (+2.8% on Cholec80, +4.3% on MICCAI2016, and +12.9% on Heichole datasets). Code will be publicly available after paper acceptance.

cs.CV↗

arXiv2025

RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills

Endowing robots with tool design abilities is critical for enabling them to solve complex manipulation tasks that would otherwise be intractable. While recent generative frameworks can automatically synthesize task settings, such as 3D scenes and reward functions, they have not yet addressed the challenge of tool-use scenarios. Simply retrieving human-designed tools might not be ideal since many tools (e.g., a rolling pin) are difficult for robotic manipulators to handle. Furthermore, existing tool design approaches either rely on predefined templates with limited parameter tuning or apply generic 3D generation methods that are not optimized for tool creation. To address these limitations, we propose RobotSmith, an automated pipeline that leverages the implicit physical knowledge embedded in vision-language models (VLMs) alongside the more accurate physics provided by physics simulations to design and use tools for robotic manipulation. Our system (1) iteratively proposes tool designs using collaborative VLM agents, (2) generates low-level robot trajectories for tool use, and (3) jointly optimizes tool geometry and usage for task performance. We evaluate our approach across a wide range of manipulation tasks involving rigid, deformable, and fluid objects. Experiments show that our method consistently outperforms strong baselines in terms of both task success rate and overall performance. Notably, our approach achieves a 50.0\% average success rate, significantly surpassing other baselines such as 3D generation (21.4%) and tool retrieval (11.1%). Finally, we deploy our system in real-world settings, demonstrating that the generated tools and their usage plans transfer effectively to physical execution, validating the practicality and generalization capabilities of our approach.

cs.RO↗

arXiv2025

Flex: End-to-End Text-Instructed Visual Navigation from Foundation Model Features

End-to-end learning directly maps sensory inputs to actions, creating highly integrated and efficient policies for complex robotics tasks. However, such models often struggle to generalize beyond their training scenarios, limiting adaptability to new environments, tasks, and concepts. In this work, we investigate the minimal data requirements and architectural adaptations necessary to achieve robust closed-loop performance with vision-based control policies under unseen text instructions and visual distribution shifts. Our findings are synthesized in Flex (Fly lexically), a framework that uses pre-trained Vision Language Models (VLMs) as frozen patch-wise feature extractors, generating spatially aware embeddings that integrate semantic and visual information. We demonstrate the effectiveness of this approach on a quadrotor fly-to-target task, where agents trained via behavior cloning on a small simulated dataset successfully generalize to real-world scenes with diverse novel goals and command formulations.

cs.RO↗

arXiv2025

Articulate AnyMesh: Open-Vocabulary 3D Articulated Objects Modeling

3D articulated objects modeling has long been a challenging problem, since it requires to capture both accurate surface geometries and semantically meaningful and spatially precise structures, parts, and joints. Existing methods heavily depend on training data from a limited set of handcrafted articulated object categories (e.g., cabinets and drawers), which restricts their ability to model a wide range of articulated objects in an open-vocabulary context. To address these limitations, we propose Articulate Anymesh, an automated framework that is able to convert any rigid 3D mesh into its articulated counterpart in an open-vocabulary manner. Given a 3D mesh, our framework utilizes advanced Vision-Language Models and visual prompting techniques to extract semantic information, allowing for both the segmentation of object parts and the construction of functional joints. Our experiments show that Articulate Anymesh can generate large-scale, high-quality 3D articulated objects, including tools, toys, mechanical devices, and vehicles, significantly expanding the coverage of existing 3D articulated object datasets. Additionally, we show that these generated assets can facilitate the acquisition of new articulated object manipulation skills in simulation, which can then be transferred to a real robotic system. Our Github website is https://articulate-anymesh.github.io.

cs.CV↗

arXiv2025

DataS^3: Dataset Subset Selection for Specialization

In many real-world machine learning (ML) applications (e.g. detecting broken bones in x-ray images, detecting species in camera traps), in practice models need to perform well on specific deployments (e.g. a specific hospital, a specific national park) rather than the domain broadly. However, deployments often have imbalanced, unique data distributions. Discrepancy between the training distribution and the deployment distribution can lead to suboptimal performance, highlighting the need to select deployment-specialized subsets from the available training data. We formalize dataset subset selection for specialization (DS3): given a training set drawn from a general distribution and a (potentially unlabeled) query set drawn from the desired deployment-specific distribution, the goal is to select a subset of the training data that optimizes deployment performance. We introduce DataS^3; the first dataset and benchmark designed specifically for the DS3 problem. DataS^3 encompasses diverse real-world application domains, each with a set of distinct deployments to specialize in. We conduct a comprehensive study evaluating algorithms from various families--including coresets, data filtering, and data curation--on DataS^3, and find that general-distribution methods consistently fail on deployment-specific tasks. Additionally, we demonstrate the existence of manually curated (deployment-specific) expert subsets that outperform training on all available data with accuracy gains up to 51.3 percent. Our benchmark highlights the critical role of tailored dataset curation in enhancing performance and training efficiency on deployment-specific distributions, which we posit will only become more important as global, public datasets become available across domains and ML models are deployed in the real world.

cs.LG↗

arXiv2025

LuciBot: Automated Robot Policy Learning from Generated Videos

Automatically generating training supervision for embodied tasks is crucial, as manual designing is tedious and not scalable. While prior works use large language models (LLMs) or vision-language models (VLMs) to generate rewards, these approaches are largely limited to simple tasks with well-defined rewards, such as pick-and-place. This limitation arises because LLMs struggle to interpret complex scenes compressed into text or code due to their restricted input modality, while VLM-based rewards, though better at visual perception, remain limited by their less expressive output modality. To address these challenges, we leverage the imagination capability of general-purpose video generation models. Given an initial simulation frame and a textual task description, the video generation model produces a video demonstrating task completion with correct semantics. We then extract rich supervisory signals from the generated video, including 6D object pose sequences, 2D segmentations, and estimated depth, to facilitate task learning in simulation. Our approach significantly improves supervision quality for complex embodied tasks, enabling large-scale training in simulators.

cs.CV↗

arXiv2025

Embodied Red Teaming for Auditing Robotic Foundation Models

Language-conditioned robot models have the potential to enable robots to perform a wide range of tasks based on natural language instructions. However, assessing their safety and effectiveness remains challenging because it is difficult to test all the different ways a single task can be phrased. Current benchmarks have two key limitations: they rely on a limited set of human-generated instructions, missing many challenging cases, and focus only on task performance without assessing safety, such as avoiding damage. To address these gaps, we introduce Embodied Red Teaming (ERT), a new evaluation method that generates diverse and challenging instructions to test these models. ERT uses automated red teaming techniques with Vision Language Models (VLMs) to create contextually grounded, difficult instructions. Experimental results show that state-of-the-art language-conditioned robot models fail or behave unsafely on ERT-generated instructions, underscoring the shortcomings of current benchmarks in evaluating real-world performance and safety. Code and videos are available at: https://s-karnik.github.io/embodied-red-team-project-page.

cs.RO↗

arXiv2024

UBSoft: A Simulation Platform for Robotic Skill Learning in Unbounded Soft Environments

It is desired to equip robots with the capability of interacting with various soft materials as they are ubiquitous in the real world. While physics simulations are one of the predominant methods for data collection and robot training, simulating soft materials presents considerable challenges. Specifically, it is significantly more costly than simulating rigid objects in terms of simulation speed and storage requirements. These limitations typically restrict the scope of studies on soft materials to small and bounded areas, thereby hindering the learning of skills in broader spaces. To address this issue, we introduce UBSoft, a new simulation platform designed to support unbounded soft environments for robot skill acquisition. Our platform utilizes spatially adaptive resolution scales, where simulation resolution dynamically adjusts based on proximity to active robotic agents. Our framework markedly reduces the demand for extensive storage space and computation costs required for large-scale scenarios involving soft materials. We also establish a set of benchmark tasks in our platform, including both locomotion and manipulation tasks, and conduct experiments to evaluate the efficacy of various reinforcement learning algorithms and trajectory optimization techniques, both gradient-based and sampling-based. Preliminary results indicate that sampling-based trajectory optimization generally achieves better results for obtaining one trajectory to solve the task. Additionally, we conduct experiments in real-world environments to demonstrate that advancements made in our UBSoft simulator could translate to improved robot interactions with large-scale soft material. More videos can be found at https://vis-www.cs.umass.edu/ubsoft/.

cs.RO↗

arXiv2024

Architect: Generating Vivid and Interactive 3D Scenes with Hierarchical 2D Inpainting

Creating large-scale interactive 3D environments is essential for the development of Robotics and Embodied AI research. Current methods, including manual design, procedural generation, diffusion-based scene generation, and large language model (LLM) guided scene design, are hindered by limitations such as excessive human effort, reliance on predefined rules or training datasets, and limited 3D spatial reasoning ability. Since pre-trained 2D image generative models better capture scene and object configuration than LLMs, we address these challenges by introducing Architect, a generative framework that creates complex and realistic 3D embodied environments leveraging diffusion-based 2D image inpainting. In detail, we utilize foundation visual perception models to obtain each generated object from the image and leverage pre-trained depth estimation models to lift the generated 2D image to 3D space. Our pipeline is further extended to a hierarchical and iterative inpainting process to continuously generate placement of large furniture and small objects to enrich the scene. This iterative structure brings the flexibility for our method to generate or refine scenes from various starting points, such as text, floor plans, or pre-arranged environments.

cs.CV↗

arXiv2024

Probing Multimodal LLMs as World Models for Driving

We provide a sober look at the application of Multimodal Large Language Models (MLLMs) in autonomous driving, challenging common assumptions about their ability to interpret dynamic driving scenarios. Despite advances in models like GPT-4o, their performance in complex driving environments remains largely unexplored. Our experimental study assesses various MLLMs as world models using in-car camera perspectives and reveals that while these models excel at interpreting individual images, they struggle to synthesize coherent narratives across frames, leading to considerable inaccuracies in understanding (i) ego vehicle dynamics, (ii) interactions with other road actors, (iii) trajectory planning, and (iv) open-set scene reasoning. We introduce the Eval-LLM-Drive dataset and DriveSim simulator to enhance our evaluation, highlighting gaps in current MLLM capabilities and the need for improved models in dynamic real-world environments.

cs.RO↗

arXiv2024

Toward Efficient Visual Gyroscopes: Spherical Moments, Harmonics Filtering, and Masking Techniques for Spherical Camera Applications

Unlike a traditional gyroscope, a visual gyroscope estimates camera rotation through images. The integration of omnidirectional cameras, offering a larger field of view compared to traditional RGB cameras, has proven to yield more accurate and robust results. However, challenges arise in situations that lack features, have substantial noise causing significant errors, and where certain features in the images lack sufficient strength, leading to less precise prediction results. Here, we address these challenges by introducing a novel visual gyroscope, which combines an Efficient Multi-Mask-Filter Rotation Estimator(EMMFRE) and a Learning based optimization(LbTO) to provide a more efficient and accurate rotation estimation from spherical images. Experimental results demonstrate superior performance of the proposed approach in terms of accuracy. The paper emphasizes the advantages of integrating machine learning to optimize analytical solutions, discusses limitations, and suggests directions for future research.

cs.CV↗