SpatialScout：VLM 智能体能否主动探索并构建 3D 场景地图？

AI 摘要

研究团队发布 SpatialScout 基准，用于评估视觉语言模型（VLM）智能体在主动物体级 3D 建图任务上的能力。该基准包含 80 个室内场景、3140 个目标物体，要求智能体在有限交互预算下控制相机探索、跨视角关联并估计物体中心坐标。实验显示最强模型 Claude Opus 5.5 平均重叠分数仅 41.71%，IoU 超过 0.9 的目标仅占 3.25%，表明 VLM 在构建完整精确的度量地图方面存在显著差距。

正文节选

SpatialScout: Can VLM Agents Actively Explore and Map 3D Scenes? Abstract SpatialScout evaluates vision-language agents on active 3D object mapping, revealing major gaps in precise metric map construction. Spatial intelligence requires agents to actively acquire visual evidence and integrate observations into a coherent representation of the three-dimensional world. Yet spatial question answering offers limited evidence of whether models can construct accurate metric maps through autonomous expl

发布时间：—

抓取时间：2026-10-22 22:16

来源机构：Hugging Face