AeroMap3D: GNSS-Denied UAV Localization

AeroMap3D aligns monocular UAV images to satellite imagery, DEM terrain, and OpenStreetMap semantics, then fuses delayed map poses for GNSS-denied 6-DoF localization.

TL;DR — AeroMap3D is a monocular 6-DoF UAV localization system for GNSS-denied navigation. According to the abstract, it aligns UAV imagery to visual, geometric, and semantic map priors; estimates scale ratio and yaw offset before dense matching; lifts 2D UAV-map correspondences onto DEM terrain; rejects semantically unreliable matches with OpenStreetMap annotations; estimates pose with RANSAC-PnP; and fuses delayed map poses with relative-motion priors using a delayed-state EKF. The abstract reports all trajectories across eight Austin sites localized within 50 m and 5.88 m mean 3D error over 55 km of flight, without UAV-Terra3D retraining or tuning.

Problem and contribution

Evidence quote: “We present AeroMap3D, a monocular 6-DoF UAV localization system that anchors onboard imagery to visual, geometric, and semantic map priors for GNSS-denied navigation.”

AeroMap3D is presented as a map-referenced aerial localization system, not merely a visual matcher. Its input is onboard monocular UAV imagery, and its references are map priors: visual, geometric, and semantic.

Evidence quote: the abstract says it addresses “two fundamental challenges in map-referenced aerial localization”: cross-view discrepancy between UAV imagery and satellite maps, and structural inconsistency between bare-earth DEMs and urban scenes. That means the contribution is best understood as an integrated pipeline for turning map-aligned visual correspondences into 6-DoF pose estimates under GNSS-denied conditions.

Full analysis, extracted claims, numerical results, entity graph, FAQ, related work, applications, and BibTeX are available via x402 micropayment.