像鸽子一样主动探索：通过智能视觉语言模型强化空间推理
Active Exploring like a Pigeon: Reinforcing Spatial Reasoning via Agentic Vision-Language Models
AI总结：
提出一种受鸽子认知地图启发的智能视觉语言模型管道，通过动态认知地图和空间断言代码提供密集奖励信号，在MindCube基准上实现80.5%的总体准确率，在Rotation子集上相对提升53.2%。
摘要：
使视觉语言模型（VLM）能够进行空间推理仍然具有挑战性。现有方法将VLM视为被动观察者，这在实际应用中难以奏效。此外，强化学习方法依赖稀疏奖励，限制了其在复杂推理任务中的有效性。受鸽子构建和利用认知地图进行导航的启发，我们提出了一种新颖的智能管道用于空间推理。首先，我们引入了一种新的\emph{动态认知地图}，将场景布局参数化为物体位置和朝向，作为新观测的持久记忆。其次，我们提出了一种新颖的\emph{空间断言代码（SAC）}，即用Python表达式编程描述空间关系。通过与动态认知地图协作，SAC能够验证中间推理步骤，提供密集的奖励信号。我们通过监督学习和强化微调来优化模型。在MindCube基准上的实验表明，我们的方法达到了\emph{80.5\%}的总体准确率，在具有挑战性的 extsc{Rotation}子集上，比当前最佳方法高出\emph{29.5}个准确率点（相对提升\emph{53.2\%}）。我们的代码和数据已在https://github.com/dw-dengwei/active-spatial-reasoning.git开源。
英文摘要：
Enabling Vision-Language Models (VLMs) to perform spatial reasoning remains challenging. Existing approaches treat VLMs as passive observers, which is difficult for real-world applications. Moreover, reinforcement learning methods rely on sparse rewards, limiting their effectiveness for complex reasoning tasks. Inspired by pigeons' building and exploiting cognitive maps for navigation, we propose a novel agentic pipeline for spatial reasoning. First, we introduce a new \emph{dynamic cognitive map} parameterizing scene layout as object positions and orientations, serving as persistent memory for new observations. Second, we propose a novel \emph{Spatial Assertion Codes (SAC)}, Python expressions programmatically describing spatial relationships. By collaborating with the dynamic cognitive map, SAC enables verification of intermediate reasoning steps, providing dense reward signals. We optimize the model via supervised and reinforcement finetuning. Experiments on the MindCube benchmark demonstrate state-of-the-art performance with \emph{80.5\%} overall accuracy, outperforming the best current method by \emph{29.5} accuracy points (a relative improvement of \emph{53.2\%}) on the challenging \textsc{Rotation} subset. Our code and data are open-sourced at https://github.com/dw-dengwei/active-spatial-reasoning.git.
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。
