具身導航器：透過指向、思考、記憶與對齊實現高效導航

摘要

儘管大型視覺-語言模型（VLMs）已顯著推動具身導航的發展，其直接部署仍具挑戰性，因為現有方法往往迫使大型視覺-語言模型進入與其二維預訓練先驗不一致的非自然動作空間，再加上僵化的推理排程與低效率的記憶管理。為克服這些限制，我們提出 TAMP-Nav，一個用於高效具身導航的統一框架。首先，我們引入像素到三維動作公式化（Point），將導航重新建構為二維視覺提示。具體而言，大型視覺-語言模型僅需選擇二維像素，再將其投影至三維座標以供底層 SLAM 控制器使用。此設計自然地將具身執行與大型視覺-語言模型固有的二維視覺能力對齊。其次，我們提出整合式選擇性推理與錨定軌跡記憶機制（Think and Memorize），該機制動態觸發鏈式思考，並僅在關鍵節點保留高保真記憶，將冗餘軌跡壓縮為輕量級時空指示器，從而保留關鍵歷史資訊並增強時空感知。最後，我們設計了一個透過群體相對策略優化（GRPO）的高效兩層對齊範式（Align）。透過疊加全局結果獎勵與細粒度過程獎勵，此密集監督將智能體的認知規劃與物理環境回饋緊密對齊，賦予模型自適應推理能力。實驗表明，TAMP-Nav 達到了最先進的效能（例如在 R2R-CE 上成功率達 66.2%），且具有高運行效率和樣本效率（僅需 90k 訓練軌跡）。

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).