ViSA-Enhanced Aerial VLN: A Visual-Spatial Reasoning Enhanced Framework for Aerial Vision-Language Navigation

TL;DR

This paper introduces ViSA, a visual-spatial reasoning framework that enhances aerial vision-language navigation by enabling direct reasoning on images, significantly improving success rates without extensive training.

Contribution

The paper presents a novel ViSA-enhanced architecture that improves spatial reasoning in aerial VLN by leveraging structured visual prompts and direct image reasoning.

Findings

70.3% success rate improvement on CityNav benchmark

Enables direct reasoning on image planes without extra training

Outperforms state-of-the-art methods significantly

Abstract

Existing aerial Vision-Language Navigation (VLN) methods predominantly adopt a detection-and-planning pipeline, which converts open-vocabulary detections into discrete textual scene graphs. These approaches are plagued by inadequate spatial reasoning capabilities and inherent linguistic ambiguities. To address these bottlenecks, we propose a Visual-Spatial Reasoning (ViSA) enhanced framework for aerial VLN. Specifically, a triple-phase collaborative architecture is designed to leverage structured visual prompting, enabling Vision-Language Models (VLMs) to perform direct reasoning on image planes without the need for additional training or complex intermediate representations. Comprehensive evaluations on the CityNav benchmark demonstrate that the ViSA-enhanced VLN achieves a 70.3\% improvement in success rate compared to the fully trained state-of-the-art (SOTA) method, elucidating its great potential as a backbone for aerial VLN systems.