面向开放世界空中目标导航的记忆增强场景理解与探索
Memory-Augmented Scene Understanding and Exploration for Open-World Aerial Object-Goal Navigation
把 RGB-D 观测增量聚合成自适应八叉树层级 3D 记忆，并用指令调制的 query 从中抽取场景/前沿 token，这套「3D 结构化记忆 + 语言引导检索」可直接迁移到任何需要长程空间记忆的视觉导航或具身任务中。
原文依据 an Adaptive Octree Memory that incrementally aggregates RGB-D observations into a hierarchical 3D representation
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论空中目标导航（Aerial ObjectNav）要求无人机仅依靠视觉观测与高层物体描述，在无逐步指令的情况下于大规模户外环境中导航至目标物体。现有方法依赖局部观测或短期历史，缺乏全面的场景理解与高效的空间探索策略，从而限制了其在复杂空中场景中的导航能力。为应对这些挑战，我们提出 OctMem-Agent，一个用于空中目标导航的八叉树记忆增强框架。具体而言，我们引入自适应八叉树记忆，将 RGB-D 观测增量式地聚合为层级化三维表示，从而在大规模空中环境中同时刻画已探索区域与未探索前沿。我们进一步提出指令引导的记忆查询模块，通过指令调制的查询抽取与任务相关的场景 token 与探索 token。通过将这些 token 与视觉观测和语言指令相结合，OctoMem-Agent 实现了全面的场景理解与有效的空间探索，以完成目标定位。在 Aerial ObjectNav 基准 UAV-ON 上的大量实验表明，本方法相较现有方法在成功率上取得 7.5% 的显著提升，验证了设计的有效性。
英文原文摘要 1429 字符
Aerial object-goal navigation (Aerial ObjectNav) requires an Unmanned Aerial Vehicle (UAV) to navigate to target objects in large-scale outdoor environments using only visual observations and high-level object descriptions, without detailed step-by-step instructions. Existing approaches rely on local observations or short-term history, lacking comprehensive scene understanding and efficient spatial exploration strategies, which constrains their navigation capability in complex aerial scenarios. To address these challenges, we propose OctMem-Agent, an octree memory-augmented framework for aerial object-goal navigation. Specifically, we introduce an Adaptive Octree Memory that incrementally aggregates RGB-D observations into a hierarchical 3D representation, capturing both explored regions and unexplored frontiers across large-scale aerial environments. We further propose a Instruction-Guided Memory Query module that extracts task-relevant scene and exploration tokens through instruction-modulated queries. By integrating these tokens with visual observations and language instructions, OctoMem-Agent achieves comprehensive scene understanding and effective spatial exploration for target localization. Extensive experiments on the Aerial ObjectNav benchmark UAV-ON demonstrate that our method achieves a significant 7.5% improvement in success rate over existing methods, validating the effectiveness of our design.