Beyond Matching to Tiles: Bridging Unaligned Aerial and Satellite Views for Vision-Only UAV Navigation

2026-03-23

AI Summary

This work proposes Bearing-UAV, a purely vision-based cross-view navigation method that jointly predicts the absolute position and heading of a drone, explicitly modeling spatial relationships to overcome the limitations of conventional map-tile matching approaches. Existing methods face a trade-off between localization accuracy and storage overhead while neglecting heading information, rendering them ineffective under large viewpoint discrepancies and misalignment common in real-world scenarios. In contrast, Bearing-UAV integrates global and local structural features and leverages relative spatial encoding for end-to-end localization without requiring pre-stored map tiles. Evaluated on Bearing-UAV-90k—a newly curated multi-city benchmark with diverse terrains—the method significantly outperforms current matching- or retrieval-based approaches, enabling lightweight and robust navigation in GNSS-denied environments.

Technology Category

• Intelligent Robots: Localization, Mapping, and Navigation
• Computer Vision: Multi-modal Vision
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Recent advances in cross-view geo-localization (CVGL) methods have shown strong potential for supporting unmanned aerial vehicle (UAV) navigation in GNSS-denied environments. However, existing work predominantly focuses on matching UAV views to onboard map tiles, which introduces an inherent trade-off between accuracy and storage overhead, and overlooks the importance of the UAV's heading during navigation. Moreover, the substantial discrepancies and varying overlaps in cross-view scenarios have been insufficiently considered, limiting their generalization to real-world scenarios. In this paper, we present Bearing-UAV, a purely vision-driven cross-view navigation method that jointly predicts UAV absolute location and heading from neighboring features, enabling accurate, lightweight, and robust navigation in the wild. Our method leverages global and local structural features and explicitly encodes relative spatial relationships, making it robust to cross-view variations, misalignment, and feature-sparse conditions. We also present Bearing-UAV-90k, a multi-city benchmark for evaluating cross-view localization and navigation. Extensive experiments show encouraging results that Bearing-UAV yields lower localization error than previous matching/retrieval paradigm across diverse terrains. Our code and dataset will be made publicly available.

Problem

• cross-view geo-localization
• UAV navigation
• GNSS-denied environments
• view misalignment
• heading estimation

Innovation

• cross-view geo-localization
• vision-only navigation
• UAV heading estimation
• spatial relationship encoding
• GNSS-denied navigation