arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.10322cs.AI
收藏

基于用户反馈的视觉-语言导航适应

User-Feedback-Driven Adaptation for Vision-and-Language Navigation

Mohamed bin Zayed University of Artificial Intelligence(Mohamed bin Zayed大学人工智能学院);School of Electrical Engineering and Computer Science, The University of Queensland(电气工程与计算机科学学院,昆士兰大学);University of Science and Technology of China(中国科学技术大学);School of Computer Science, The University of Adelaide(计算机科学学院,阿德莱德大学)

Yongqiang Yu, Xuhui Li, Hazza Mahmood, Jinxing Zhou, Haodong Hong, Longtao Jiang, Zhiqiang Xu, Qi Wu, Xiaojun Chang

AI总结:

本文提出基于用户反馈的视觉-语言导航适应方法,通过拓扑感知轨迹构建和持久记忆库机制,提升代理对用户指令的响应能力,实现高效模仿学习和跨指令风格的强适应性。

摘要:

视觉-语言导航(VLN)代理在现实世界中的部署受到离线训练后可靠监督稀缺的限制。尽管最近的适应方法试图通过环境驱动的自监督(例如熵最小化)来缓解分布偏移,但这些信号往往嘈杂,并可能导致代理在长时间序列决策中放大其自身的错误。在本文中,我们提出了一种范式转变,将用户反馈,特别是回合级的成功确认和目标级的纠正,作为VLN的主要且通用的监督信号。与内部置信度分数不同,用户反馈是意图对齐且现场一致的,直接纠正代理与用户指令的脱节。为了有效利用这种监督,我们引入了用户反馈驱动的学习框架,该框架包含拓扑感知的轨迹构建流程。这种机制通过在代理逐步构建的拓扑图上生成可行路径,将稀疏的目标级纠正提升为密集的路径级监督,使样本高效模仿学习成为可能,而无需需要逐步的人类示范。此外,我们开发了一种持久记忆库机制用于预热初始化,支持在导航会话中重用之前获得的拓扑和缓存的表示。在GSA-R2R基准上的大量实验表明,我们的方法将稀疏交互转化为稳健的监督,一致地优于环境驱动的基线,同时在多样化的指令风格中表现出强大的适应性。

英文摘要:

Real-world deployment of Vision-and-Language Navigation (VLN) agents is constrained by the scarcity of reliable supervision after offline training. While recent adaptation methods attempt to mitigate distribution shifts via environment-driven self-supervision (e.g., entropy minimization), these signals are often noisy and can cause the agent to amplify its own mistakes during long-horizon sequential decision-making. In this paper, we propose a paradigm shift that positions user feedback, specifically episode-level success confirmations and goal-level corrections, as a primary and general-purpose supervision signal for VLN. Unlike internal confidence scores, user feedback is intent-aligned and in-situ consistent, directly correcting the agent's decoupling from user instructions. To effectively leverage this supervision, we introduce a user-feedback-driven learning framework featuring a topology-aware trajectory construction pipeline. This mechanism lifts sparse, goal-level corrections into dense path-level supervision by generating feasible paths on the agent's incrementally built topological graph, enabling sample-efficient imitation learning without requiring step-by-step human demonstrations. Furthermore, we develop a persistent memory bank mechanism for warm-start initialization, supporting the reuse of previously acquired topology and cached representations across navigation sessions. Extensive experiments on the GSA-R2R benchmark demonstrate that our approach transforms sparse interaction into robust supervision, consistently outperforming environment-driven baselines while exhibiting strong adaptability across diverse instruction styles.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑