Unmanned Autonomous Aerial Navigation in GPS-Denied Environments

Abstract

This work focuses on enhancing existing technology to support unmanned aerial vehicle navigation for the inspection of power plants. These inspections are a vital component of a power plant's capability to function but can be costly and dangerous to do by humans. Therefore, it has been proposed to utilize a small, unmanned quadcopter integrated with autonomous navigation by using an alternate signal in place of GPS. The quadcopter can gain the ability to navigate to desired coordinates without the need for GPS, which can often be unavailable within these structures. The UAV uses an optical flow sensor to determine the ground velocity and x-y position, and sonar for measuring the altitude. Using ROS and MAVLink, the communication can be modified to use "fake GPS" data, instead of the actual GPS signal, to be published in the appropriate ROS topics to avoid GPS failures. The optical flow and fake GPS are filtered by an Extended Kalman Filter to get a better position estimation during navigation. Through this process, autonomous flight was successfully achieved without GPS.