MarvisNav: Making Memory Visible on Route Choices for Zero-Shot Object Navigation

📄 arXiv: 2610.06510v1 📥 PDF

作者: Jincheng Wang, Chi Pui Chan, Wei Zeng, Shuyang Zhang, Jianhao Jiao, Dimitrios Kanoulas

分类: cs.RO

发布日期: 2026-10-05

备注: 22 pages

🔗 代码/项目: PROJECT_PAGE

💡 一句话要点

提出MarvisNav以解决零-shot物体导航中的记忆可视化问题

🎯 匹配领域: 支柱六：视频提取与匹配 (Video Extraction)

关键词: 零-shot物体导航 视觉-语言模型 探索记忆 拓扑图 路径选择 机器人导航 智能决策

📋 核心要点

  1. 现有的零-shot物体导航方法将探索历史与当前视图分开表示，导致记忆与路径选择之间的对应关系隐含，影响导航效果。
  2. MarvisNav框架通过将探索记忆直接可视化在视觉路径选择中，允许视觉-语言模型同时评估目标相关性和探索状态，简化了处理流程。
  3. 在HM3D数据集上，MarvisNav在没有策略训练的情况下实现了81.2%的成功率，且在实际机器人实验中表现出良好的可部署性。

📝 摘要（中文）

在寻找物体时，人们会考虑可能的目标位置和已探索的地方。现有的零-shot物体导航方法中，视觉-语言模型通常将探索历史与当前视图分开表示，导致记忆与路径选择之间的对应关系隐含。本文提出MarvisNav框架，使探索记忆在视觉路径选择中直接可视化，通过维护拓扑图并将候选节点及其探索状态投影到自我中心视图中，MarvisNav能够让视觉-语言模型共同评估目标相关性和探索状态。实验结果表明，MarvisNav在HM3D上实现了81.2%的成功率和42.5%的路径长度比，且在MP3D上也表现出竞争力，验证了其实际可部署性。

🔬 方法详解

问题定义：本文旨在解决零-shot物体导航中探索记忆与路径选择之间的分离问题，现有方法通常将两者分开处理，导致信息丢失和效率低下。

核心思路：MarvisNav通过将探索记忆直接可视化在自我中心视图中，使得视觉-语言模型能够同时考虑目标相关性和探索状态，避免了额外的融合步骤。

技术框架：MarvisNav的整体架构包括一个拓扑图的维护模块和一个将候选节点及其探索状态投影到自我中心视图的模块，形成记忆承载的视觉路径选择。

关键创新：MarvisNav的创新在于将探索状态与视觉候选直接绑定，使得视觉-语言模型能够在单一阶段内进行评估，显著提高了效率和准确性。

关键设计：在设计中，探索状态不仅仅是二元的访问状态，而是捕捉局部探索进展，确保模型能够充分利用历史信息进行决策。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

MarvisNav在HM3D数据集上实现了81.2%的成功率和42.5%的路径长度比，且在MP3D上表现出竞争力。此外，该方法在实际机器人实验中验证了其可部署性，显示出相较于其他VLM方法显著减少了VLM调用次数（如WMNav的7.5%）。

🎯 应用场景

MarvisNav的研究成果在智能机器人导航、自动化仓储和无人驾驶等领域具有广泛的应用潜力。通过提升物体导航的效率和准确性，该方法能够在复杂环境中实现更智能的决策，推动相关技术的进步和实际应用。

📄 摘要（原文）

When searching for an object, people choose their next move by considering both likely target locations and places already explored. The current view can cue place-associated memories, bringing target relevance and prior exploration into the same spatial context. In many zero-shot object navigation (ZSON) methods, however, vision-language models (VLMs) infer promising search areas from egocentric images, while exploration history is represented separately, e.g., as text or maps. This separation either requires an additional fusion step or leaves the correspondence between memory and route choices implicit for the VLM to recover. We instead make exploration memory directly visible on visual route choices. We propose MarvisNav, a ZSON framework that maintains a topological graph and projects candidate nodes together with their exploration states onto egocentric views as memory-bearing visual route choices. These states capture local exploration progress beyond binary visitation. By binding exploration state directly to each visual candidate, MarvisNav enables the VLM to jointly evaluate target relevance and exploration state without a separate post-hoc fusion or reranking stage. Without policy training, MarvisNav achieves state-of-the-art performance on HM3D (81.2% SR and 42.5% SPL), while remaining competitive on MP3D. It also outperforms representative VLM-based methods with far fewer VLM calls (e.g., 7.5% of WMNav). Real-robot experiments across diverse scenes further validate its practical deployability. Beyond MarvisNav, our study shows that memory representation shapes VLM decisions and ZSON performance, highlighting that effective memory use depends not only on its availability, but also on how it is represented. Code and project page will be available at \url{https://wangjincheng1998.github.io/MarvisNav/}.