Publication

Autonomy Artificial Intelligence Robotics (AAIR)
2018

CROSS CORRELATING GROUND-LEVEL PANORAMAS WITH SATELLITE IMAGERY FOR GPS-DENIED LOCALIZATION OF AUTONOMOUS GROUND VEHICLES

by Calvin Cheung; Stanley Baek

Abstract

In this paper, we explore the usage of normalized cross-correlation to perform localization in GPS-denied environments. Spherical panoramic images from a ground vehicle are transformed to top-down viewpoints and compared with satellite imagery using normalized cross-correlation to find the vehicle’s location within the satellite image. The implementation of this system has yielded positive results when tested upon publicly available panoramic images and satellite imagery, with the identified locations being within an average of 46.95 meters from the ground truth.