Visual-inertial-GPS fusion for robust UAV navigation

Abstract

Unmanned aerial vehicles (UAV) have been widely used in a variety of applications, from the military, where they assist with search and rescue [1], and reconnaissance operations, to industrial applications which help to ease the inspection and maintenance processes. Along with the rise of UAVs, research on outdoor and indoor navigation techniques is gaining popularity. Currently, navigation systems are optimized for a specific environment over the other, equipping these UAVs with the ability to seamlessly transition between indoor and outdoor environments can make them significantly more versatile and useful in a wide variety of operations [2]. This can be achieved through a tightly coupled system that consists of a camera and inertial measurement unit (IMU) for visual-inertial odometry, the Global Navigation Satellite System (GNSS) for outdoor positioning, a WiFi indoor positioning system, and lastly an indoor-outdoor scene classification model.