Under Review

PEnG: Pose-Enhanced Geo-Localisation

Combines CVGL and relative pose estimation to achieve sub-1m precision, improving localisation accuracy and outperforming prior methods by reducing median distance errors by 96.9%.

WACV 2025

SpaGBOL: Spatial-Graph-Based Orientated Localisation

GNN-based approach leveraging graph-structures to enhance CVGL accuracy, achieving SOTA results with novel contributions in data, proximity-based feature exploitation, and retrieval filtering.

IROS 2024

BEV-CV: Birds-Eye-View Transform for Cross-View Geo-Localisation

Transforms ground-level images into BEV for direct comparison with aerial images and moving to realistic FOV constraints. BEV-CV achieves SOTA accuracy and reduces computational costs.

AINTEC 2022

Constrained Machine Learning for LoRa Gateway Location Optimisation

ML ensemble to enhance LoRaWAN scalability by optimizing device clustering, gateway placement, and signal strength, achieving significant improvements in distance reduction, received signal strength, and network throughput.