One Policy, Many Embodiments: Unified Camera-Centric Action Geometry Pre-training for Heterogeneous Embodied Manipulation

• URL: http://arxiv.org/abs/2608.26058v1
• Date: Wed, 26 Aug 2026 17:27:36 GMT
• Title: One Policy, Many Embodiments: Unified Camera-Centric Action Geometry Pre-training for Heterogeneous Embodied Manipulation
• Abstract summary: UCAG-P is a camera-centric unified action formulation that aligns heterogeneous embodied datasets into a shared geometric action space.<n>A geometry-conditioned action translator combines predicted motion with target-embodiment kinematics to produce executable controls.<n> UCAG-P is trained on 4.03K hours of robot and simulation data and 2.34K hours of human demonstrations.
• Score: 17.443768996704033
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Scaling generalist vision-language-action (VLA) policies is severely bottlenecked by the inherent heterogeneity of embodied data, which spans diverse robot morphologies, camera configurations, and low-level action spaces. Existing paradigms typically address this mismatch through explicit action retargeting, human-to-robot video synthesis, or dataset-specific adaptation branches, fundamentally hindering the joint learning of a unified policy. We introduce UCAG-P, a camera-centric unified action formulation that structurally aligns heterogeneous embodied datasets into a shared geometric action space. Rather than treating robot-specific commands as the shared policy target, UCAG-P represents manipulation through camera-observable anchor motion in image and camera-frame coordinates, treating robot arms, humanoids, and human hands as different embodiments of a common action schema. A geometry-conditioned action translator combines predicted motion with target-embodiment kinematics to produce executable controls. The resulting decoupled architecture allows a shared VLA policy to learn transferable manipulation geometry while retaining embodiment-specific controllability. UCAG-P is trained on 4.03K hours of robot and simulation data and 2.34K hours of human demonstrations. A single checkpoint reaches 98.3% on LIBERO, 88.7% and 89.2% on RoboTwin Easy and Hard, 82.0% zero-shot on LIBERO-Plus, and 62.0% on RoboCasa GR-1, without benchmark-specific fine-tuning.

Related papers

• AnyWorld: Factorized Egocentric World Models for Cross-Embodiment Generalization [88.25615187835321]
  We propose AnyWorld, a cross-embodiment world modeling framework.<n>Our model factorizes an interaction into action, camera, and embodiment.<n>We train the model with large-scale human interaction pretraining.
  arXiv  Detail & Related papers  (2026-08-29T12:55:15Z)
• Hydra-0: Action Flow for Generalist World Modeling and Control [64.88849745875753]
  Hydra-0 is a world model conditioned on action flow, which represents robot actions as pixel motion.<n>Our best configuration achieves 90.4% lower robot-motion error and 60.2% lower object-motion error.<n>A trained action head maps the resulting latent features to executable actions without requiring task-specific expert robot demonstrations.
  arXiv  Detail & Related papers  (2026-08-18T17:59:30Z)
• DyPES-VLA: Learning Shared Dynamics Priors and Embodiment-Specific Control for Cross-Embodiment Manipulation [18.601805553878638]
  DyPES-VLA is a cross-embodiment VLA that learns shared Dynamics Priors and Embodiment-Specific control.<n>As a generalist policy, our ourmethod achieves state-of-the-art performance across simulation and real-world evaluations.
  arXiv  Detail & Related papers  (2026-08-06T17:59:20Z)
• CORE: Common Outcome Regularities from Action-Free Visual Demonstrations for Robot Manipulation [26.48033419994886]
  CORE is a policy learning framework that extracts Common Outcome Regularities from visual demonstrations.<n> CORE exploits a key observation: although successful trajectories for the same task can be diverse, their terminal states often share stable object configurations.
  arXiv  Detail & Related papers  (2026-06-28T17:27:23Z)
• SPACE: Enabling Learning from Cross-Robot Data Toward Generalist Policies [20.583080585659513]
  We introduce State Prediction and Adaptive Command Execution (SPACE) framework.<n>SPACE handles robot dynamics variation at three levels: across different embodiments, across hardware units of the same embodiment, and within a single robot during operation.<n>Experiments show that SPACE substantially outperforms policies that directly predict control commands when learning from data collected across different embodiments and across hardware units of the same embodiment.
  arXiv  Detail & Related papers  (2026-06-23T01:32:40Z)
• Qwen-VLA: Unifying Vision-Language-Action Modeling across Tasks, Environments, and Robot Embodiments [96.23886784364997]
  Embodied intelligence is often studied through specialized models for individual tasks such as manipulation or navigation.<n>We present Qwen-VLA, a unified embodied foundation model that extends Qwen's vision-language modeling stack to continuous action and trajectory generation.
  arXiv  Detail & Related papers  (2026-05-28T17:36:31Z)
• MiVLA: Towards Generalizable Vision-Language-Action Model with Human-Robot Mutual Imitation Pre-training [102.850162490626]
  We propose MiVLA, a vision-language-action model empowered by human-robot mutual imitation pre-training.<n>We show that MiVLA achieves strong improved generalization capability, outperforming state-of-the-art VLAs.
  arXiv  Detail & Related papers  (2025-12-17T12:59:41Z)
• Opening the Sim-to-Real Door for Humanoid Pixel-to-Action Policy Transfer [59.02729900344616]
  GPU-accelerated, photorealistic simulation has opened a scalable data-generation path for robot learning.<n>We develop a teacher-student-bootstrap learning framework for vision-based humanoid loco-manipulation.<n>This represents the first humanoid sim-to-real policy capable of diverse articulated loco-manipulation using pure RGB perception.
  arXiv  Detail & Related papers  (2025-11-30T20:07:13Z)
• The One RING: a Robotic Indoor Navigation Generalist [58.30694487843546]
  RING (Robotic Indoor Navigation Generalist) is an embodiment-agnostic policy that turns any mobile robot into an effective indoor semantic navigator.<n>Trained entirely in simulation, RING leverages large-scale randomization over robot embodiments to enable robust generalization to many real-world platforms.
  arXiv  Detail & Related papers  (2024-12-18T23:15:41Z)
• IRASim: A Fine-Grained World Model for Robot Manipulation [24.591694756757278]
  We present IRASim, a novel world model capable of generating videos with fine-grained robot-object interaction details.<n>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.
  arXiv  Detail & Related papers  (2024-06-20T17:50:16Z)
• Self-Supervised Motion Retargeting with Safety Guarantee [12.325683599398564]
  We present a data-driven motion method that enables the generation of natural motions in humanoid robots from motion capture data or RGB videos. Our method can generate expressive robotic motions from both the CMU motion capture database and YouTube videos.
  arXiv  Detail & Related papers  (2021-03-11T04:17:26Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.