XR-1: Towards Versatile Vision-Language-Action Models via Learning Unified Vision-Motion Representations

Recent progress in large-scale robotic datasets and vision-language models (VLMs) has advanced research on vision-language-action (VLA) models. However, existing VLA models still face two fundamental challenges: (\textit{i}) producing precise low-level actions from high-dimensional observations, (\textit{ii}) bridging domain gaps across heterogeneous data sources, including diverse robot embodiments and human demonstrations. Existing methods often encode latent variables from either visual dynamics or robotic actions to guide policy learning, but they fail to fully exploit the complementary multi-modal knowledge present in large-scale, heterogeneous datasets. In this work, we present \textbf{XR-1}, a novel framework for versatile and scalable VLA learning across diverse robots, tasks, and environments. At its core, XR-1 introduces the \emph{Unified Vision-Motion Codes (UVMC)}, a discrete latent representation learned via a dual-branch VQ-VAE that jointly encodes visual dynamics and robotic motion. UVMC addresses these challenges by (\textit{i}) serving as an intermediate representation between the observations and actions, and (\textit{ii}) aligning multimodal dynamic information from heterogeneous data sources to capture complementary knowledge. To effectively exploit UVMC, we propose a \emph{three-stage training paradigm}: (\textit{i}) self-supervised UVMC learning, (\textit{ii}) UVMC-guided pretraining on large-scale cross-embodiment robotic datasets, and (\textit{iii}) task-specific post-training. We validate XR-1 through extensive real-world experiments with more than 12,000 rollouts on six different robot embodiments, spanning over 120 diverse manipulation tasks. XR-1 consistently outperforms state-of-the-art baselines such as $\pi_0$ and GR00T-N1.5 while demonstrating strong generalization to novel objects, background variations, distractors, and illumination changes. Our project is at \href{https://xr-1-vla.github.io/}{https://xr-1-vla.github.io/}.

XR-1: A General Robot Model That Connects Seeing and Acting Across Different Robots AI systems are now very good at chatting, writing, and creating images, but robots still struggle in the physical world. A robot may see a complex scene, yet fail to turn that visual information into precise fingertip movements, and it often cannot learn a new skill simply by watching a video as humans do. To address this challenge, we developed XR-1, a general robot model designed to work across different robot bodies. At its core is UVMC, a shared “visual-action language” that acts like a translator between what a robot sees and how it moves. UVMC converts both visual motion and physical actions into a unified digital representation, allowing XR-1 to bridge differences in robot hardware and even learn useful action patterns from human demonstration videos. We tested XR-1 in 14,000 real-world trials across 6 robot platforms and more than 120 tasks, including bimanual pouring and precise door opening. XR-1 outperformed leading existing models and remained robust in unfamiliar cluttered scenes and changing lighting conditions. These results bring us closer to building general-purpose robot assistants that can adapt to many tasks, environments, and bodies.