Enhancement of a commercial multicopter for research in autonomous navigation

Abstract

Multicopters are lightweight and maneuverable aerial vehicles yet unable to carry heavy payloads, such as large sensors or computers required for indoor autonomous navigation. Therefore, localization is usually performed by using vision-based solutions employing of either lightweight on-board cameras or external fixed cameras and a ground station for data-processing. Nevertheless, the current tendency is to use a low-power on-board computers to perform all computation on the multicopter itself. This paper covers the enhancement of a commercial multicopter, also called drone, with computation ability and sensorial devices for autonomous flight without the need of a ground-station. We describe the hardware and software integrated into the drone, which will be used for the future development of 6DoF navigation algorithms. The resulting system is able to work with most standard sensors and has the possibility to change them as needed. Also, we demonstrate the correct behavior of the drone by using a test navigation program that autonomously follows a moving beacon at constant distance and controlled altitude using an RGB-D camera and a sonar.