Visual Embodied Brain: Let Multimodal Large Language Models See, Think, and Control in Spaces

2025-05-30

AI Summary

Existing approaches struggle to unify multimodal understanding, visual-spatial reasoning, and physical interaction capabilities. This paper introduces VeBrain—the first unified vision-language-action framework for embodied intelligence—formulating robot control as text-based tasks grounded in 2D visual space. Our key contributions are: (1) a novel textual spatial control paradigm that grounds language instructions in pixel-aligned visual coordinates; (2) a lightweight robot adapter enabling end-to-end mapping from LLM-generated commands to low-level motion policies; and (3) VeBrain-600k, a high-quality embodied instruction dataset comprising 600K samples with multimodal chain-of-thought annotations. VeBrain achieves state-of-the-art performance across 13 cross-modal and 5 spatial reasoning benchmarks, outperforming Qwen2.5-VL by 5.6% on MMVet and improving average success rates by 50% on quadruped robot tasks. Real-world deployment demonstrates strong generalization and compositional zero-shot capability.

Technology Category

• Intelligent Robots: Embodied AI
• Computer Vision: Multi-modal Vision
• Natural Language Processing: Language Grounding & Multi-modal NLP

Application Category

• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Economics, Online Markets and Human Computation: LLM based quality controls for crowd work

Abstract

The remarkable progress of Multimodal Large Language Models (MLLMs) has attracted increasing attention to extend them to physical entities like legged robot. This typically requires MLLMs to not only grasp multimodal understanding abilities, but also integrate visual-spatial reasoning and physical interaction capabilities. Nevertheless,existing methods struggle to unify these capabilities due to their fundamental differences.In this paper, we present the Visual Embodied Brain (VeBrain), a unified framework for perception, reasoning, and control in real world. VeBrain reformulates robotic control into common text-based MLLM tasks in the 2D visual space, thus unifying the objectives and mapping spaces of different tasks. Then, a novel robotic adapter is proposed to convert textual control signals from MLLMs to motion policies of real robots. From the data perspective, we further introduce VeBrain-600k, a high-quality instruction dataset encompassing various capabilities of VeBrain. In VeBrain-600k, we take hundreds of hours to collect, curate and annotate the data, and adopt multimodal chain-of-thought(CoT) to mix the different capabilities into a single conversation. Extensive experiments on 13 multimodal benchmarks and 5 spatial intelligence benchmarks demonstrate the superior performance of VeBrain to existing MLLMs like Qwen2.5-VL. When deployed to legged robots and robotic arms, VeBrain shows strong adaptability, flexibility, and compositional capabilities compared to existing methods. For example, compared to Qwen2.5-VL, VeBrain not only achieves substantial gains on MMVet by +5.6%, but also excels in legged robot tasks with +50% average gains.

Problem

• Extend MLLMs to physical entities like legged robots
• Unify visual-spatial reasoning and physical interaction capabilities
• Convert textual control signals to real robot motion policies

Innovation

• Unified framework for perception, reasoning, control
• Text-based MLLM tasks in 2D visual space
• Robotic adapter converts text to motion policies