AdaNav: Adaptive Reasoning with Uncertainty for Vision-Language Navigation
University of Science and Technology of China · Microsoft · Microsoft Research · State Key Laboratory of Novel Software Technology, Nanjing University · Central South University · nanjing university · Microsoft Research Asia · Tsinghua University · Institute for AI Industry Research (AIR), Tsinghua University · Tsinghua University, Tsinghua University
PDF 由论文原始站点提供，PaperCompass 不保存论文文件。
摘要
Vision-Language Navigation (VLN) requires agents to follow natural language instructions by grounding them in sequential visual observations over long horizons. Explicit reasoning could enhance temporal consistency and perception–action alignment, but reasoning at fixed steps often leads to suboptimal performance and unnecessary computation. To address this, we propose AdaNav, an uncertainty-based adaptive reasoning framework for VLN. At its core is the Uncertainty-Adaptive Reasoning Block (UAR), a lightweight plugin that dynamically triggers reasoning. We introduce Action Entropy as a policy prior for UAR and progressively refine it through a Heuristics-to-RL training method, enabling agents to learn difficulty-aware reasoning policies under the strict data limitations of embodied tasks. Results show that with only 6K training samples, AdaNav achieves substantial gains over closed-source models trained on million-scale data, improving success rate by 20% on R2R val-unseen, 11.7% on RxR-CE, and 11.4% in real-world scenes.