Understanding and predicting the dynamics of the physical world can enhance a robot's ability to plan and interact effectively in complex environments. While recent video generation models have shown strong potential in modeling dynamic scenes, generating videos that are both temporally coherent and geometrically consistent across camera views remains a significant challenge. To address this, we propose a 4D video generation model that enforces multi-view 3D consistency of videos by supervising the model with cross-view pointmap alignment during training. This geometric supervision enables the model to learn a shared 3D representation of the scene, allowing it to predict future video sequences from novel viewpoints based solely on the given RGB-D observations, without requiring camera poses as inputs. Compared to existing baselines, our method produces more visually stable and spatially aligned predictions across multiple simulated and real-world robotic datasets. We further show that the predicted 4D videos can be used to recover robot end-effector trajectories using an off-the-shelf 6DoF pose tracker, supporting robust robot manipulation and generalization to novel camera viewpoints.
Recent advances in vision-language-action (VLA) models have shown promise in integrating image generation with action prediction to improve generalization and reasoning in robot manipulation. However, existing methods are limited to challenging image-based forecasting, which suffers from redundant information and lacks comprehensive and critical world knowledge, including dynamic, spatial and semantic information. To address these limitations, we propose DreamVLA, a novel VLA framework that integrates comprehensive world knowledge forecasting to enable inverse dynamics modeling, thereby establishing a perception-prediction-action loop for manipulation tasks. Specifically, DreamVLA introduces a dynamic-region-guided world knowledge prediction, integrated with the spatial and semantic cues, which provide compact yet comprehensive representations for action planning. This design aligns with how humans interact with the world by first forming abstract multimodal reasoning chains before acting. To mitigate interference among the dynamic, spatial and semantic information during training, we adopt a block-wise structured attention mechanism that masks their mutual attention, preventing information leakage and keeping each representation clean and disentangled. Moreover, to model the conditional distribution over future actions, we employ a diffusion-based transformer that disentangles action representations from shared latent features. Extensive experiments on both real-world and simulation environments demonstrate that DreamVLA achieves 76.7% success rate on real robot tasks and 4.44 average length on the CALVIN ABC-D benchmarks.
Imitation learning (IL) aims to enable robots to perform tasks autonomously by observing a few human demonstrations. Recently, a variant of IL, called In-Context IL, utilized off-the-shelf large language models (LLMs) as instant policies that understand the context from a few given demonstrations to perform a new task, rather than explicitly updating network models with large-scale demonstrations. However, its reliability in the robotics domain is undermined by hallucination issues such as LLM-based instant policy, which occasionally generates poor trajectories that deviate from the given demonstrations. To alleviate this problem, we propose a new robust in-context imitation learning algorithm called the robust instant policy (RIP), which utilizes a Student's t-regression model to be robust against the hallucinated trajectories of instant policies to allow reliable trajectory generation. Specifically, RIP generates several candidate robot trajectories to complete a given task from an LLM and aggregates them using the Student's t-distribution, which is beneficial for ignoring outliers (i.e., hallucinations); thereby, a robust trajectory against hallucinations is generated. Our experiments, conducted in both simulated and real-world environments, show that RIP significantly outperforms state-of-the-art IL methods, with at least $26\%$ improvement in task success rates, particularly in low-data scenarios for everyday tasks. Video results available at https://sites.google.com/view/robustinstantpolicy.
Vision-Language-Action (VLA) models have emerged as a prominent research area, showcasing significant potential across a variety of applications. However, their performance in endoscopy robotics, particularly endoscopy capsule robots that perform actions within the digestive system, remains unexplored. The integration of VLA models into endoscopy robots allows more intuitive and efficient interactions between human operators and medical devices, improving both diagnostic accuracy and treatment outcomes. In this work, we design CapsDT, a Diffusion Transformer model for capsule robot manipulation in the stomach. By processing interleaved visual inputs, and textual instructions, CapsDT can infer corresponding robotic control signals to facilitate endoscopy tasks. In addition, we developed a capsule endoscopy robot system, a capsule robot controlled by a robotic arm-held magnet, addressing different levels of four endoscopy tasks and creating corresponding capsule robot datasets within the stomach simulator. Comprehensive evaluations on various robotic tasks indicate that CapsDT can serve as a robust vision-language generalist, achieving state-of-the-art performance in various levels of endoscopy tasks while achieving a 26.25% success rate in real-world simulation manipulation.
Preference-based Reinforcement Learning (PbRL) methods provide a solution to avoid reward engineering by learning reward models based on human preferences. However, poor feedback- and sample- efficiency still remain the problems that hinder the application of PbRL. In this paper, we present a novel efficient query selection and preference-guided exploration method, called SENIOR, which could select the meaningful and easy-to-comparison behavior segment pairs to improve human feedback-efficiency and accelerate policy learning with the designed preference-guided intrinsic rewards. Our key idea is twofold: (1) We designed a Motion-Distinction-based Selection scheme (MDS). It selects segment pairs with apparent motion and different directions through kernel density estimation of states, which is more task-related and easy for human preference labeling; (2) We proposed a novel preference-guided exploration method (PGE). It encourages the exploration towards the states with high preference and low visits and continuously guides the agent achieving the valuable samples. The synergy between the two mechanisms could significantly accelerate the progress of reward and policy learning. Our experiments show that SENIOR outperforms other five existing methods in both human feedback-efficiency and policy convergence speed on six complex robot manipulation tasks from simulation and four real-worlds.
Imitation learning from large multi-task demonstration datasets has emerged as a promising path for building generally-capable robots. As a result, 1000s of hours have been spent on building such large-scale datasets around the globe. Despite the continuous growth of such efforts, we still lack a systematic understanding of what data should be collected to improve the utility of a robotics dataset and facilitate downstream policy learning. In this work, we conduct a large-scale dataset composition study to answer this question. We develop a data generation framework to procedurally emulate common sources of diversity in existing datasets (such as sensor placements and object types and arrangements), and use it to generate large-scale robot datasets with controlled compositions, enabling a suite of dataset composition studies that would be prohibitively expensive in the real world. We focus on two practical settings: (1) what types of diversity should be emphasized when future researchers collect large-scale datasets for robotics, and (2) how should current practitioners retrieve relevant demonstrations from existing datasets to maximize downstream policy performance on tasks of interest. Our study yields several critical insights -- for example, we find that camera poses and spatial arrangements are crucial dimensions for both diversity in collection and alignment in retrieval. In real-world robot learning settings, we find that not only do our insights from simulation carry over, but our retrieval strategies on existing datasets such as DROID allow us to consistently outperform existing training strategies by up to 70%. More results at https://robo-mimiclabs.github.io/
This work demonstrates how autonomously learning aspects of robotic operation from sparsely-labeled, real-world data of deployed, engineered solutions at industrial scale can provide with solutions that achieve improved performance. Specifically, it focuses on multi-suction robot picking and performs a comprehensive study on the application of multi-modal visual encoders for predicting the success of candidate robotic picks. Picking diverse items from unstructured piles is an important and challenging task for robot manipulation in real-world settings, such as warehouses. Methods for picking from clutter must work for an open set of items while simultaneously meeting latency constraints to achieve high throughput. The demonstrated approach utilizes multiple input modalities, such as RGB, depth and semantic segmentation, to estimate the quality of candidate multi-suction picks. The strategy is trained from real-world item picking data, with a combination of multimodal pretrain and finetune. The manuscript provides comprehensive experimental evaluation performed over a large item-picking dataset, an item-picking dataset targeted to include partial occlusions, and a package-picking dataset, which focuses on containers, such as boxes and envelopes, instead of unpackaged items. The evaluation measures performance for different item configurations, pick scenes, and object types. Ablations help to understand the effects of in-domain pretraining, the impact of different modalities and the importance of finetuning. These ablations reveal both the importance of training over multiple modalities but also the ability of models to learn during pretraining the relationship between modalities so that during finetuning and inference, only a subset of them can be used as input.
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/
Manipulation has long been a challenging task for robots, while humans can effortlessly perform complex interactions with objects, such as hanging a cup on the mug rack. A key reason is the lack of a large and uniform dataset for teaching robots manipulation skills. Current robot datasets often record robot action in different action spaces within a simple scene. This hinders the robot to learn a unified and robust action representation for different robots within diverse scenes. Observing how humans understand a manipulation task, we find that understanding how the objects should move in the 3D space is a critical clue for guiding actions. This clue is embodiment-agnostic and suitable for both humans and different robots. Motivated by this, we aim to learn a 3D flow world model from both human and robot manipulation data. This model predicts the future movement of the interacting objects in 3D space, guiding action planning for manipulation. Specifically, we synthesize a large-scale 3D optical flow dataset, named ManiFlow-110k, through a moving object auto-detect pipeline. A video diffusion-based world model then learns manipulation physics from these data, generating 3D optical flow trajectories conditioned on language instructions. With the generated 3D object optical flow, we propose a flow-guided rendering mechanism, which renders the predicted final state and leverages GPT-4o to assess whether the predicted flow aligns with the task description. This equips the robot with a closed-loop planning ability. Finally, we consider the predicted 3D optical flow as constraints for an optimization policy to determine a chunk of robot actions for manipulation. Extensive experiments demonstrate strong generalization across diverse robotic manipulation tasks and reliable cross-embodiment adaptation without hardware-specific training.
We introduce OG-VLA, a novel architecture and learning framework that combines the generalization strengths of Vision Language Action models (VLAs) with the robustness of 3D-aware policies. We address the challenge of mapping natural language instructions and multi-view RGBD observations to quasi-static robot actions. 3D-aware robot policies achieve state-of-the-art performance on precise robot manipulation tasks, but struggle with generalization to unseen instructions, scenes, and objects. On the other hand, VLAs excel at generalizing across instructions and scenes, but can be sensitive to camera and robot pose variations. We leverage prior knowledge embedded in language and vision foundation models to improve generalization of 3D-aware keyframe policies. OG-VLA projects input observations from diverse views into a point cloud which is then rendered from canonical orthographic views, ensuring input view invariance and consistency between input and output spaces. These canonical views are processed with a vision backbone, a Large Language Model (LLM), and an image diffusion model to generate images that encode the next position and orientation of the end-effector on the input scene. Evaluations on the Arnold and Colosseum benchmarks demonstrate state-of-the-art generalization to unseen environments, with over 40% relative improvements while maintaining robust performance in seen settings. We also show real-world adaption in 3 to 5 demonstrations along with strong generalization. Videos and resources at https://og-vla.github.io/
Robot manipulation, especially bimanual manipulation, often requires setting up multiple cameras on multiple robot manipulators. Before robot manipulators can generate motion or even build representations of their environments, the cameras rigidly mounted to the robot need to be calibrated. Camera calibration is a cumbersome process involving collecting a set of images, with each capturing a pre-determined marker. In this work, we introduce the Bi-Manual Joint Calibration and Representation Framework (Bi-JCR). Bi-JCR enables multiple robot manipulators, each with cameras mounted, to circumvent taking images of calibration markers. By leveraging 3D foundation models for dense, marker-free multi-view correspondence, Bi-JCR jointly estimates: (i) the extrinsic transformation from each camera to its end-effector, (ii) the inter-arm relative poses between manipulators, and (iii) a unified, scale-consistent 3D representation of the shared workspace, all from the same captured RGB image sets. The representation, jointly constructed from images captured by cameras on both manipulators, lives in a common coordinate frame and supports collision checking and semantic segmentation to facilitate downstream bimanual coordination tasks. We empirically evaluate the robustness of Bi-JCR on a variety of tabletop environments, and demonstrate its applicability on a variety of downstream tasks.
Fine-grained robot manipulation, such as lifting and rotating a bottle to display the label on the cap, requires robust reasoning about object parts and their relationships with intended tasks. Despite recent advances in training general-purpose robot manipulation policies guided by language instructions, there is a notable lack of large-scale datasets for fine-grained manipulation tasks with part-level instructions and diverse 3D object instances annotated with part-level labels. In this work, we introduce PartInstruct, the first large-scale benchmark for training and evaluating fine-grained robot manipulation models using part-level instructions. PartInstruct comprises 513 object instances across 14 categories, each annotated with part-level information, and 1302 fine-grained manipulation tasks organized into 16 task classes. Our training set consists of over 10,000 expert demonstrations synthesized in a 3D simulator, where each demonstration is paired with a high-level task instruction, a chain of base part-based skill instructions, and ground-truth 3D information about the object and its parts. Additionally, we designed a comprehensive test suite to evaluate the generalizability of learned policies across new states, objects, and tasks. We evaluated several state-of-the-art robot manipulation approaches, including end-to-end vision-language policy learning and bi-level planning models for robot manipulation on our benchmark. The experimental results reveal that current models struggle to robustly ground part concepts and predict actions in 3D space, and face challenges when manipulating object parts in long-horizon tasks.
The field of robotics has made significant strides toward developing generalist robot manipulation policies. However, evaluating these policies in real-world scenarios remains time-consuming and challenging, particularly as the number of tasks scales and environmental conditions change. In this work, we demonstrate that world models can serve as a scalable, reproducible, and reliable proxy for real-world robot policy evaluation. A key challenge is generating accurate policy videos from world models that faithfully reflect the robot actions. We observe that directly inputting robot actions or using high-dimensional encoding methods often fails to generate action-following videos. To address this, we propose Policy2Vec, a simple yet effective approach to turn a video generation model into a world simulator that follows latent action to generate the robot video. We then introduce WorldEval, an automated pipeline designed to evaluate real-world robot policies entirely online. WorldEval effectively ranks various robot policies and individual checkpoints within a single policy, and functions as a safety detector to prevent dangerous actions by newly developed robot models. Through comprehensive paired evaluations of manipulation policies in real-world environments, we demonstrate a strong correlation between policy performance in WorldEval and real-world scenarios. Furthermore, our method significantly outperforms popular methods such as real-to-sim approach.
This paper evaluates single-view mesh reconstruction models for creating digital twin environments in robot manipulation. Recent advances in computer vision for 3D reconstruction from single viewpoints present a potential breakthrough for efficiently creating virtual replicas of physical environments for robotics contexts. However, their suitability for physics simulations and robotics applications remains unexplored. We establish benchmarking criteria for 3D reconstruction in robotics contexts, including handling typical inputs, producing collision-free and stable reconstructions, managing occlusions, and meeting computational constraints. Our empirical evaluation using realistic robotics datasets shows that despite success on computer vision benchmarks, existing approaches fail to meet robotics-specific requirements. We quantitively examine limitations of single-view reconstruction for practical robotics implementation, in contrast to prior work that focuses on multi-view approaches. Our findings highlight critical gaps between computer vision advances and robotics needs, guiding future research at this intersection.
A key challenge in robot manipulation lies in developing policy models with strong spatial understanding, the ability to reason about 3D geometry, object relations, and robot embodiment. Existing methods often fall short: 3D point cloud models lack semantic abstraction, while 2D image encoders struggle with spatial reasoning. To address this, we propose SEM (Spatial Enhanced Manipulation model), a novel diffusion-based policy framework that explicitly enhances spatial understanding from two complementary perspectives. A spatial enhancer augments visual representations with 3D geometric context, while a robot state encoder captures embodiment-aware structure through graphbased modeling of joint dependencies. By integrating these modules, SEM significantly improves spatial understanding, leading to robust and generalizable manipulation across diverse tasks that outperform existing baselines.
Robot manipulation learning from human demonstrations offers a rapid means to acquire skills but often lacks generalization across diverse scenes and object placements. This limitation hinders real-world applications, particularly in complex tasks requiring dexterous manipulation. Vision-Language-Action (VLA) paradigm leverages large-scale data to enhance generalization. However, due to data scarcity, VLA's performance remains limited. In this work, we introduce Object-Focus Actor (OFA), a novel, data-efficient approach for generalized dexterous manipulation. OFA exploits the consistent end trajectories observed in dexterous manipulation tasks, allowing for efficient policy training. Our method employs a hierarchical pipeline: object perception and pose estimation, pre-manipulation pose arrival and OFA policy execution. This process ensures that the manipulation is focused and efficient, even in varied backgrounds and positional layout. Comprehensive real-world experiments across seven tasks demonstrate that OFA significantly outperforms baseline methods in both positional and background generalization tests. Notably, OFA achieves robust performance with only 10 demonstrations, highlighting its data efficiency.
Vision-Language Models (VLMs) acquire real-world knowledge and general reasoning ability through Internet-scale image-text corpora. They can augment robotic systems with scene understanding and task planning, and assist visuomotor policies that are trained on robot trajectory data. We explore the reverse paradigm - using rich, real, multi-modal robot trajectory data to enhance and evaluate VLMs. In this paper, we present Robo2VLM, a Visual Question Answering (VQA) dataset generation framework for VLMs. Given a human tele-operated robot trajectory, Robo2VLM derives ground-truth from non-visual and non-descriptive sensory modalities, such as end-effector pose, gripper aperture, and force sensing. Based on these modalities, it segments the robot trajectory into a sequence of manipulation phases. At each phase, Robo2VLM uses scene and interaction understanding to identify 3D properties of the robot, task goal, and the target object. The properties are used to generate representative VQA queries - images with textural multiple-choice questions - based on spatial, goal-conditioned, and interaction reasoning question templates. We curate Robo2VLM-1, a large-scale in-the-wild dataset with 684,710 questions covering 463 distinct scenes and 3,396 robotic manipulation tasks from 176k real robot trajectories. Results suggest that Robo2VLM-1 can benchmark and improve VLM capabilities in spatial and interaction reasoning.
World models, which predict transitions based on history observation and action sequences, have shown great promise in improving data efficiency for sequential decision making. However, existing world models often require extensive domain-specific training and still produce low-fidelity, coarse predictions, limiting their applicability in complex environments. In contrast, video diffusion models trained on large, internet-scale datasets have demonstrated impressive capabilities in generating high-quality videos that capture diverse real-world dynamics. In this work, we present Vid2World, a general approach for leveraging and transferring pre-trained video diffusion models into interactive world models. To bridge the gap, Vid2World performs casualization of a pre-trained video diffusion model by crafting its architecture and training objective to enable autoregressive generation. Furthermore, it introduces a causal action guidance mechanism to enhance action controllability in the resulting interactive world model. Extensive experiments in robot manipulation and game simulation domains show that our method offers a scalable and effective approach for repurposing highly capable video diffusion models to interactive world models.
Large Language Models (LLMs) enable planning from natural language instructions using implicit world knowledge, but often produce flawed plans that require refinement. Instead of directly predicting plans, recent methods aim to learn a problem domain that can be solved for different goal states using classical planners. However, these approaches require significant human feedback to obtain useful models. We address this shortcoming by learning hierarchical domains, where low-level predicates and actions are composed into higher-level counterparts, and by leveraging simulation to validate their preconditions and effects. This hierarchical approach is particularly powerful for long-horizon planning, where LLM-based planning approaches typically struggle. Furthermore, we introduce a central error reasoner to ensure consistency among the different planning levels. Evaluation on two challenging International Planning Competition (IPC) domains and a long-horizon robot manipulation task demonstrates higher planning success rates than state-of-the-art domain synthesis and LLM-modulo planning methods, while constructing high-quality models of the domain. Resources, videos and detailed experiment results are available at https://claudius-kienle.github.io/lodge/.
Training vision-based manipulation policies that are robust across diverse visual environments remains an important and unresolved challenge in robot learning. Current approaches often sidestep the problem by relying on invariant representations such as point clouds and depth, or by brute-forcing generalization through visual domain randomization and/or large, visually diverse datasets. Disentangled representation learning - especially when combined with principles of associative memory - has recently shown promise in enabling vision-based reinforcement learning policies to be robust to visual distribution shifts. However, these techniques have largely been constrained to simpler benchmarks and toy environments. In this work, we scale disentangled representation learning and associative memory to more visually and dynamically complex manipulation tasks and demonstrate zero-shot adaptability to visual perturbations in both simulation and on real hardware. We further extend this approach to imitation learning, specifically Diffusion Policy, and empirically show significant gains in visual generalization compared to state-of-the-art imitation learning methods. Finally, we introduce a novel technique adapted from the model equivariance literature that transforms any trained neural network policy into one invariant to 2D planar rotations, making our policy not only visually robust but also resilient to certain camera perturbations. We believe that this work marks a significant step towards manipulation policies that are not only adaptable out of the box, but also robust to the complexities and dynamical nature of real-world deployment. Supplementary videos are available at https://sites.google.com/view/vis-gen-robotics/home.