CLOSER-VLN: Closed-Loop Self-Verified Retrieval-Augmented Reasoning for Aerial Vision-Language Navigation

2026-06-24

AI Summary

This work addresses the limitations of existing vision-and-language navigation methods, which predominantly rely on open-loop mechanisms that lack validation and correction of intermediate actions, leading to error accumulation and frequent deviations from the target in aerial navigation. To mitigate this, the paper introduces a closed-loop self-verification mechanism—a training-free, retrieval-augmented reasoning framework. Before executing each action, the framework performs sequential action reasoning followed by multidimensional reliability verification; if verification fails, it dynamically triggers multimodal retrieval to refine the planned action. Evaluated on unseen environments in CityNav, the approach achieves a success rate (SR) of 32.01% and a path-length-weighted success rate (SPL) of 21.28%, substantially demonstrating the efficacy of closed-loop reasoning in vision-and-language navigation.

Abstract

Vision-language navigation (VLN) has recently advanced with large language and multimodal models, enabling agents to follow natural-language instructions in unseen environments without training a task-specific navigation policy. However, most existing VLN methods relying on large models still adopt an open-loop decision-execution approach, where candidate actions are generated from instructions and observations but are rarely verified or corrected before execution. This causes critical issues in aerial VLN, where minor errors in intermediate actions may quickly accumulate into large trajectory deviations and lead to target loss. To address this issue, we propose Closed-loop Self-verified Retrieval-augmented Reasoning (CLOSER), a training-policy-free method that sequentially performs action reasoning, reliability verification, targeted retrieval, and action correction in a closed-loop manner before executing concrete actions. We instantiate the CLOSER in aerial VLN tasks and develop a CLOSER-VLN framework, which is composed of three components: a hierarchical reasoner for generating candidate actions based on available information, a multidimensional action verifier for assessing the reliability of actions generated by the reasoner, and a verification-triggered multimodal retriever for retrieving targeted exemplars from a memory bank only when verification fails. We conduct experimental evaluations on the CityNav benchmark, where CLOSER-VLN achieves 32.01% SR and 21.28% SPL on the test-unseen split, confirming the effectiveness of closed-loop reasoning.

Problem

• Vision-Language Navigation
• Aerial Navigation
• Closed-Loop Reasoning
• Action Verification
• Trajectory Deviation

Innovation

• closed-loop reasoning
• self-verification
• retrieval-augmented navigation
• aerial vision-language navigation
• training-policy-free