Visual simultaneous localization and mapping for co-operating unmanned aerial vehicles

This paper presents a modified open-source Red Green Blue - Depth Simultaneous Localization and Mapping (RGB-D SLAM) that addresses the problem of multi-robot SLAM in unmanned aerial vehicles (UAVs). The modified RGB-D SLAM system was tested with quadrotors which were equipped with RGB-D cameras as the only sensor for solving the SLAM problem. Two quadrotors were tasked to autonomously explore an unknown indoor environment using a 2D navigation stack while still producing a 3D map using RGB-D SLAM. The paper also describes the implementation of a control interface where the Roll- Pitch-Yaw-Thrust (RPYT) channels of a commercially available flight controller can be controlled to achieve autonomous flight control. Height and yaw position control were also implemented. © 2015, Mechatronics and Machine Vision in Practice. All rights reserved