arXiv 2511.074032026-07-07cs.CVcs.AIcs.CLcs.LG

空间思考者：通过密集奖励强化场景图基础的空间推理

SpatialThinker: Reinforcing Scene Graph-Grounded Spatial Reasoning via Dense Rewards

University of Oxford（牛津大学）；University of California, Santa Cruz（加州大学圣克鲁兹分校）

Hunar Batra, Haoqin Tu, Hardy Chen, Yuanze Lin, Cihang Xie, Ronald Clark

备注：: Preprint. Accepted at NeurIPS 2025 Workshops on SPACE in Vision, Language, and Embodied AI (SpaVLE) as Oral, Embodied World Models for Decision Making (EWM), Aligning Reinforcement Learning Experimentalists and Theorists (ARLET), and Scaling Environments for Agents (SEA)

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

研究针对多模态大语言模型空间推理难题，提出SpatialThinker，通过在线强化学习统一场景图生成与视觉推理，构建心理场景图并借助密集奖励推理，贡献包括基于SGG推理、高质量训练数据集及密集奖励设计。

摘要：

多模态大语言模型在视觉语言任务中取得进展，但空间推理仍有困难。现有空间MLLMs存在不足。我们引入SpatialThinker，首次通过在线RL统一场景图生成和视觉推理。模型通过构建心理场景图模拟人类空间感知并通过密集空间奖励推理。贡献有三：基于SGG推理、STVQA - 7K数据集、密集空间奖励设计。SpatialThinker - 7B相比稀疏RL有更大提升。

英文摘要：

Multimodal large language models (MLLMs) have achieved remarkable progress in vision-language tasks, but continue to struggle with spatial reasoning. Existing spatial MLLMs rely on large-scale datasets, explicit 3D inputs, architecture-specific modifications, or sparse Reinforcement Learning (RL) methods that provide insufficient guidance for spatially-grounded reasoning. We introduce SpatialThinker. To our knowledge, it is the first MLLM unifying Scene Graph Generation (SGG) and visual reasoning in a single pass via online RL. The model simulates human-like spatial perception by constructing a mental scene graph of task-relevant objects and relations, and reasoning toward an answer via dense spatial rewards. Our contributions are threefold: (1) SGG-grounded reasoning: integrating SGG directly within the reasoning chain rather than as a disjoint preprocessing step; (2) STVQA-7K: a high-quality spatial VQA training dataset via a scalable synthesis pipeline; and (3) a dense spatial reward design that enforces structured grounding during RL and generalizes to improve broad visual perception. SpatialThinker-7B achieves 3.6$\times$ larger gains over SFT and $1.7\times$ better in- and out-of-distribution generalization than sparse RL. Trained on only 7K samples, SpatialThinker-7B matches GPT-5 and outperforms GPT-4o, while SpatialThinker-30B surpasses both GPT-5 and Claude 4 Sonnet on average across 14 spatial and real-world benchmarks, demonstrating that structured spatial grounding with reward-aligned reasoning enables robust spatial understanding with limited data.

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。