Reinforcing Dual-Path Reasoning in Spatial Vision Language Models

📄 arXiv: 2606.17539v1 📥 PDF

作者: Yatai Ji, An-Chieh Cheng, Yang Fu, Yukang Chen, Han Zhang, Zhaojing Yang, Wei Huang, Ka Chun Cheung, Song Han, Vidya Nariyambut Murali, Pavlo Molchanov, Jan Kautz, Simon See, Hongxu Yin, Ping Luo, Sifei Liu

分类: cs.CV, cs.AI

发布日期: 2026-06-16


💡 一句话要点

提出SR-REAL以解决复杂空间推理问题

🎯 匹配领域: 支柱二:RL算法与架构 (RL & Architecture) 支柱九:具身大模型 (Embodied Foundation Models)

关键词: 空间推理 视觉语言模型 强化学习 几何感知 多模态学习 3D检测 模型优化

📋 核心要点

  1. 现有空间视觉语言模型在处理复杂空间推理时面临多步推理的挑战,尤其是在深度、距离和场景关系方面。
  2. 本文提出的SR-REAL框架结合了语言推理和3D几何检测两条互补的推理路径,以应对不同类型的空间查询。
  3. SR-REAL在多项空间基准测试中表现优异,显示出在区域感知任务中的3D定位精度和一般空间推理能力的提升。

📝 摘要(中文)

空间视觉语言模型在几何感知方面取得了显著进展,但复杂的空间推理仍然具有挑战性。不同的空间查询需要不同的策略,本文提出了双路径空间推理框架SR-REAL,结合了语言推理和3D检测。通过强化学习优化模型,SR-REAL在多种空间基准测试中显著超越了现有模型,展示了良好的泛化能力和任务间的正向迁移。

🔬 方法详解

问题定义:本文旨在解决空间视觉语言模型在复杂空间推理中面临的多步推理挑战,现有方法在处理不同类型空间查询时缺乏灵活性和准确性。

核心思路:SR-REAL框架通过引入语言推理(LOR)和检测后推理(DTR)两条路径,分别处理语言推理和3D几何信息,从而实现更高效的空间推理。

技术框架:SR-REAL的整体架构包括两个主要模块:首先是冷启动的监督微调阶段,构建LOR和DTR的思维链监督;其次是通过强化学习优化模型,提升推理准确性和格式奖励。

关键创新:SR-REAL的最大创新在于同时训练两条推理路径,促进相互强化,且通过高质量的混合冷启动数据实现稳定的强化学习优化,这在现有方法中尚属首次。

关键设计:模型设计中采用了离散中心检测奖励来进一步优化几何对齐,损失函数和奖励机制的设置确保了模型在不同任务间的良好迁移和泛化能力。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在多项空间基准测试中,SR-REAL显著超越了现有空间视觉语言模型,尤其是在区域感知任务中,DTR路径通过精确的3D定位实现了性能提升,整体模型在不同数据集和领域间表现出良好的泛化能力,未进行每任务调优的情况下实现了正向迁移。

🎯 应用场景

该研究的潜在应用领域包括机器人导航、增强现实和自动驾驶等需要复杂空间推理的场景。通过提高空间视觉语言模型的推理能力,SR-REAL能够在多种实际应用中提供更准确的环境理解和决策支持,未来可能推动智能系统的广泛应用。

📄 摘要(原文)

Spatial VLMs have made substantial progress in geometric perception, yet complex spatial reasoning requiring multi-step inference over depth, distance, and scene relations remains challenging. Moreover, different spatial queries call for fundamentally different strategies: some are best addressed through purely linguistic, step-by-step deduction, while others require explicit 3D grounding before quantitative inference. We present Dual-Path Spatial Reasoning via Reinforcement Learning for Spatial VLMs (SR-REAL), a unified framework that equips a spatial VLM with two complementary reasoning paths: Language-Only Reasoning (LOR), which performs step-by-step linguistic deduction, and Detect-Then-Reason (DTR), which detects 3D geometric cues (e.g., centers or bounding boxes) via region tokens before explicit geometric inference. SR-REAL begins with a cold-start supervised fine-tuning stage that constructs LOR and DTR chain-of-thought supervision and exposes a region-to-3D interface, followed by RL that optimizes the policy model with accuracy and format rewards; for DTR, a discrete center-based detection reward further refines geometric alignment. Across diverse spatial benchmarks, SR-REAL significantly outperforms spatial VLM baselines: (i) a single RL-trained model supports both reasoning paths, with DTR excelling in region-aware tasks through precise 3D localization and LOR enhancing general spatial reasoning; (ii) jointly training both paths fosters mutual reinforcement; (iii) high-quality, blended cold-start data is crucial for stable RL optimization; and (iv) the model generalizes across datasets and domains without per-task tuning, demonstrating positive transfer between LOR and DTR.