Capek 0.5: An Execution-Centric Vision-Language Model for Embodied Intelligence

📄 arXiv: 2608.06756v1 📥 PDF

作者: Ying Chen, Weizhen Li, Zhe Hu, Zhenjiang Li, Rui Jiang, Zhifeng Gu, Lihuang Fang, Jiangping Liu, Lei Yi, Jie Chen

分类: cs.AI

发布日期: 2026-08-07

💡 一句话要点

提出Capek 0.5以解决机器人执行过程中的多模态理解问题

🎯 匹配领域: 支柱二：RL算法与架构 (RL & Architecture)

关键词: 具身智能 视觉-语言模型 能力分类 强化学习 多模态理解 机器人执行 状态验证 动态环境

📋 核心要点

  1. 现有方法通常针对特定任务开发能力，缺乏对执行过程整体的组织与整合，导致多模态理解能力不足。
  2. 论文提出Capek 0.5，通过执行中心能力分类法整合空间推理、时间理解、动作指导和状态验证四种能力，提升机器人执行效率。
  3. Capek 0.5在综合基准测试中表现优异，保留所有专业能力，并在闭环环境中成功执行任务，显示出显著的性能提升。

📝 摘要（中文）

随着视觉-语言模型在具身智能体中的应用日益增多，机器人执行过程具有迭代性，每个动作都会重塑场景和物理状态。现有方法通常针对孤立的任务目标开发能力，缺乏围绕执行整体的组织与整合。本文提出Capek 0.5，一个基于执行中心能力分类法的具身视觉-语言模型。该模型通过强化学习获取四个能力家族：空间推理、时间理解、动作指导和状态验证，并通过权重空间合并和策略空间蒸馏将这些能力整合为一个推理模型。Capek 0.5在多个基准测试中表现优异，保留了所有四种专业能力，并成功转移至闭环具身任务执行。

🔬 方法详解

问题定义：本文旨在解决现有视觉-语言模型在机器人执行过程中能力整合不足的问题。现有方法往往针对特定任务开发，缺乏对执行整体的考虑，导致多模态理解能力的不足。

核心思路：Capek 0.5通过执行中心能力分类法，将具身能力按功能角色进行分类，形成四个能力家族，旨在提升机器人在动态环境中的执行能力。

技术框架：模型的整体架构包括四个主要模块：空间推理、时间理解、动作指导和状态验证。每个模块通过强化学习获得，并在推理时通过权重合并和策略蒸馏整合为一个统一模型。

关键创新：Capek 0.5的创新在于其能力分类法的设计，强调了能力在执行过程中的功能角色，而非单纯的任务导向。这种设计使得模型在多模态理解和执行任务时更加高效。

关键设计：模型采用共享骨干网络进行能力学习，使用可验证的奖励信号进行强化学习。通过权重空间合并和策略空间蒸馏，确保各个能力模块在一个检查点中得以保留，且量化损失得以控制。

📊 实验亮点

Capek 0.5在多个基准测试中表现出色，较初始化状态在大多数基准行上都有显著提升。所有四种专业能力在一个检查点中得以保留，且在闭环具身任务执行中成功转移，展示了其在实际应用中的有效性。

🎯 应用场景

Capek 0.5的研究成果在机器人领域具有广泛的应用潜力，尤其是在动态环境中的任务执行，如自动驾驶、服务机器人和工业自动化等。其多模态理解能力的提升将促进机器人在复杂场景中的自主决策和执行能力，推动具身智能的发展。

📄 摘要（原文）

Vision-language models are increasingly serving as the reasoning core of embodied agents. Robot execution is inherently iterative: each action reshapes the scene and physical state, continually renewing what must be perceived, reasoned about, and verified. Meeting these demands requires complementary capabilities that differ in supervision signals, prediction formats, and verification criteria. Existing approaches typically develop these capabilities against isolated, task-specific objectives, leaving open how they should be organized and integrated around execution as a whole. We present Capek 0.5, an embodied vision-language model built around an execution-centric capability taxonomy. Rather than organizing training by datasets or tasks, the taxonomy groups embodied capabilities according to their functional roles throughout execution and comprises four capability families: Spatial Reasoning, Temporal Understanding, Action Guidance, and State Verification. Each capability is first acquired by a dedicated specialist through reinforcement learning with verifiable rewards from a shared backbone, and the specialists are then consolidated into a single inference-time model through weight-space merging followed by routed policy-space distillation. We instantiate Capek 0.5 at the 2B and 35B-A3B scales and evaluate it from three complementary perspectives: comprehensive benchmark suites including Capek-StateBench, a new benchmark for state verification; a controlled study of capability retention from specialists to the unified model; and closed-loop evaluation in simulated embodied environments. Capek 0.5 improves the large majority of matched benchmark rows over its initialization, retains all four specialized capabilities in one checkpoint with quantified losses, and transfers to closed-loop embodied task execution.