NavHarness: Towards Lifelong Embodied Navigation
作者: Xunyi Zhao, Jian Zhou, Sihao Lin, Gengze Zhou, Zerui Li, Xinyu Yan, Jiajun Liu, Anton van den Hengel, Qi Wu
分类: cs.RO
发布日期: 2026-09-28
💡 一句话要点
提出NavHarness以解决长期嵌入式导航中的记忆处理问题
🎯 匹配领域: 支柱九：具身大模型 (Embodied Foundation Models)
关键词: 长期导航 嵌入式系统 记忆处理 多轮推理 机器人导航 任务成功率 智能代理
📋 核心要点
- 现有方法在连续嵌入式导航任务中面临记忆处理不足的问题，导致导航效率低下。
- NavHarness通过将记忆处理整合到导航循环中，利用多轮会话来校对和记录经验，从而提升导航能力。
- 在GOAT-Bench上，NavHarness在任务成功率上分别提高了18.6和22.6分，展示了其在长期导航中的有效性。
📝 摘要（中文）
前沿模型在单一嵌入式导航任务中表现良好，但在连续任务中，代理需要依赖不断演变的地图和早期搜索记录，这些信息可能不完整或与新观察相冲突。本文提出NavHarness，一个无训练的嵌入式导航工具，使记忆处理成为导航循环的一部分。它通过多轮代理会话利用地图、任务记录和知识，校对观察结果并记录修正，以指导行动。NavHarness在GOAT-Bench上显著提高了任务成功率，表明经验的有效传递对长期导航至关重要。
🔬 方法详解
问题定义：本文旨在解决长期嵌入式导航中，代理在连续任务中如何有效利用记忆的问题。现有方法往往无法处理不完整或冲突的信息，导致导航性能下降。
核心思路：NavHarness的核心思路是将记忆处理融入导航循环，通过多轮会话利用历史信息来指导当前任务的执行。这种设计使得代理能够在新任务中有效利用先前的经验。
技术框架：NavHarness的整体架构包括多个模块：地图管理、任务记录、知识库和观察校对。代理在导航过程中会不断更新这些模块，以确保信息的准确性和完整性。
关键创新：NavHarness的最大创新在于其训练自由的设计，使得记忆处理成为导航的核心部分，显著提升了代理在新任务中的适应能力。与传统方法相比，它能够更好地整合和利用历史经验。
关键设计：在技术细节上，NavHarness采用了结构化的恢复交接方式，优于传统的长度匹配摘要。此外，使用SLAM估计的姿态和GPT-6 Astra的结合，进一步提升了任务成功率。具体参数设置和损失函数设计在论文中详细阐述。
🖼️ 关键图片
📊 实验亮点
NavHarness在GOAT-Bench上实现了83.7的状态成功率和36.9的有效成功率，相较于传统方法，分别提高了18.6和22.6分，显示出其在长期导航任务中的显著优势。
🎯 应用场景
NavHarness的研究成果在智能家居、服务机器人和自动驾驶等领域具有广泛的应用潜力。通过提升代理的长期导航能力，可以实现更智能的环境交互和任务执行，推动机器人技术的进步与普及。
📄 摘要（原文）
Frontier models can now perform well on individual embodied navigation tasks through multi-round multimodal reasoning with simple tools. Across successive tasks, however, an agent must also rely on an evolving map and earlier search records, both of which may be incomplete or conflict with new observations. We present NavHarness, a training-free embodied harness towards lifelong navigation that makes memory processing part of the navigation loop. During navigation, its multi-round agentic session draws on maps, task records, and house knowledge, checking them against observations and recording corrections to guide its actions. NavHarness preserves this experience across fresh conversations for new tasks or recovery attempts, while outcome verification and run-end summaries support its later reuse. On GOAT-Bench, NavHarness improves s-SR over context-only independent sessions by 18.6 points with Astra and 22.6 with Opus 5. Using SLAM-estimated poses, NavHarness with GPT-6 Astra achieves state-of-the-art task success of 83.7 s-SR with 36.9 e-SR on GOAT-Bench and 85.9 s-SR on IR2R-CE. To understand these gains, we examine how experience is carried between sessions and find that structured recovery handovers outperform length-matched summaries. In extended deployments across houses, consolidation improves navigation beyond retaining maps and task records, with case studies showing how agents use earlier experience to interpret new goals, investigate unresolved questions, and resume failed searches. We suggest that progress towards lifelong navigation depends on how successive reasoning sessions build on prior experience, alongside improvements in single-task capability.