RoboPIN: 基于锚定思维链的具身推理
RoboPIN: Grounded Embodied Reasoning via Pinned Chain-of-Thought
AI总结：
提出Pinned Chain-of-Thought (PinCoT)推理范式，通过结构化视觉锚点绑定实体，解决多步推理中实体引用漂移和视觉解耦问题；训练4B参数模型在14个基准上平均超越最强7B基线12%。
摘要：
具身推理要求模型感知物理环境中与任务相关的物体和空间，并在多步推理中保持一致的视觉基础。然而，当前的视觉语言模型依赖于纯文本或坐标增强的思维链，其中实体引用仍然隐式和模糊。这可能导致推理过程与视觉证据解耦、实体引用在步骤间漂移、推理轨迹与最终答案之间的因果断裂，并且由于跨视角外观变化，这些问题在多视角场景中进一步放大。为了解决这些问题，我们提出了Pinned Chain-of-Thought (PinCoT)，一种结构化推理范式，将每个推理步骤锚定到视觉证据。PinCoT引入了推理锚点的概念，它将每个任务相关实体绑定到一个结构化的视觉锚点，包含实体名称、唯一标识、视角索引和空间基础，从而能够在推理步骤和视角之间实现一致的实体跟踪。我们构建了一个全自动数据生成管道来构建数据集，这是一个高质量的PinCoT格式推理数据集。然后，我们通过三阶段后训练训练方法，逐步注入具身知识、结构化推理能力和过程监督对齐，奖励直接约束推理过程中的锚点定位和身份一致性。在涵盖具身空间推理、多视角推理和指向的14个基准测试中，仅有4B参数的方法始终优于7B级别的开源具身模型，比最强的7B基线Mimo-Embodied平均提高12%。进一步分析表明，PinCoT提高了基础准确性和跨步骤身份一致性，验证了过程监督的有效性。
英文摘要：
Embodied reasoning requires models to perceive task-relevant objects and spaces in physical environments and maintain consistent visual grounding throughout multi-step reasoning. However, current vision-language models rely on text-only or coordinate-augmented chain-of-thought, where entity references remain implicit and ambiguous. This may cause the reasoning process to decouple from visual evidence, entity references to drift across steps, and a causal disconnection between the reasoning trajectory and the final answer, with these problems further amplified in multi-view scenarios due to cross-view appearance changes. To address these issues, we propose Pinned Chain-of-Thought (PinCoT), a structured reasoning paradigm that pins every reasoning step to visual evidence. PinCoT introduces the concept of reasoning anchor, which binds each task-relevant entity to a structured visual anchor with entity name, unique identity, view index, and spatial grounding, enabling consistent entity tracking across reasoning steps and views. We build a fully automated data generation pipeline to construct PIN-170K, a high-quality PinCoT-formatted reasoning dataset. We then train RoboPIN through three-stage post-training that progressively injects embodied knowledge, structured reasoning ability, and process-supervised alignment, with rewards that directly constrain both anchor localization and identity consistency during reasoning. On 14 benchmarks covering embodied spatial reasoning, multi-view reasoning, and pointing, RoboPIN with only 4B parameters surpasses 7B level open-source embodied models on average, achieving a 12% average improvement over the strongest 7B baseline, Mimo-Embodied. Further analysis shows that PinCoT improves grounding accuracy and cross-step identity consistency, validating the effectiveness of process supervision.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
