MAV based SLAM and autonomous navigation: A view towards efficient on-board systems

In the field of autonomous drone or micro air vehicle (MAV) research, much of the existing literature focuses on novel approaches to MAV automation and navigation. Whilst discovering these new approaches has scientific merit, these works rarely focus on the impact that the deployment of such systems have in terms of the operational time, power consumption or efficiency of the MAV. This work sets out to review the parallel tracking and mapping algorithm (PTAM) as applied to MAV control systems. Through experimentation, the limits of this algorithm are found in an attempt to determine the minimum computational and power requirements for a computer to have, in order to run PTAM effectively. This work demonstrates that it is feasible with current available technology, to operate PTAM on a 5 watt computer by limiting the parameters that add computational overhead to the system.