具身导航器：指向、思考、记忆与对齐实现高效导航

摘要

尽管大型视觉语言模型（VLMs）显著推进了具身导航的发展，但其直接部署仍具挑战性，因为现有方法常常迫使VLMs进入与其二维预训练先验不一致的非自然动作空间，加之僵化的推理调度和低效的记忆管理。为克服这些局限，我们提出了TAMP-Nav，一个用于高效具身导航的统一框架。首先，我们引入了一种像素到三维动作公式化方法（Point），将导航重新表述为二维视觉提示。具体而言，VLM仅需选择二维像素，随后这些像素被投影到三维坐标中，供底层SLAM控制器使用。这种设计自然地使具身执行与VLM固有的二维视觉能力对齐。其次，我们提出了一种集成的选择性推理与锚定轨迹记忆机制（Think and Memorize），该机制动态触发思维链，并仅在关键节点保留高保真记忆，将冗余轨迹压缩为轻量级时空指示符，从而保留关键历史信息并增强时空感知。最后，我们通过群体相对策略优化（GRPO）设计了一种高效的两级对齐范式（Align）。通过将全局结果奖励与细粒度过程奖励叠加，这种密集监督使智能体的认知规划与物理环境反馈紧密对齐，赋予模型自适应推理能力。实验表明，TAMP-Nav达到了最先进的性能（例如在R2R-CE上取得66.2%的SR），并具有较高的运行效率和样本效率（仅需9万条训练轨迹）。

English

Although Large Vision-Language Models (VLMs) have significantly advanced embodied navigation, their direct deployment remains challenging, as existing methods often force VLMs into unnatural action spaces that misalign with their 2D pre-training priors, compounded by rigid reasoning schedules and inefficient memory management. To overcome these limitations, we propose TAMP-Nav, a unified framework for efficient embodied navigation. First, we introduce a Pixel-to-3D Action Formulation (Point) that reformulates navigation into 2D visual prompting. Specifically, the VLM merely selects 2D pixels, which are then projected into 3D coordinates for a low-level SLAM controller. This design naturally aligns embodied execution with the VLM's inherent 2D visual capabilities. Second, we propose an integrated Selective Reasoning and Anchor-Trajectory Memory mechanism (Think and Memorize), which dynamically triggers Chain-of-Thought and retains high-fidelity memory only at critical nodes, compressing redundant trajectories into lightweight Space-Time Indicators, thereby preserving critical historical information and enhancing spatio-temporal perception. Finally, we design an efficient Two-Level Alignment Paradigm (Align) via Group Relative Policy Optimization (GRPO). By superimposing global outcome rewards with fine-grained process rewards, this dense supervision tightly aligns the agent's cognitive planning with physical environmental feedback, endowing the model with adaptive reasoning capabilities. Experiments demonstrate that TAMP-Nav achieves state-of-the-art performance (e.g., 66.2% SR on R2R-CE) with high runtime and sample efficiency (requiring only 90k training trajectories).