A Quadrotor UAV Navigational Command and Control Aid: A Landing Pad Detection and Localisation System

Research and development of quadrotor UAV robots is a fast-growing area. There is currently a high R&D focus by many companies on UAVs for delivery purposes, swarm applications, farming applications, search and rescue missions, environmental studies and other advanced applications. GPS is a navigational system that can be used to guide a UAV toward a target destination. However, as GPS navigation is prone to local errors, there is a vital need for another sensor to assist with short-range navigation and finding the target landing position. In this case, the quadrotor's on-board camera could act as a useful, optical, sensor to collect real-time data about a destination area and the location of a landing pad. In this research work, successful landing pad detection was accomplished by employing the most ideal of a number of classifiers, which were trained using OpenCV. Two different types of classifier training were performed: these were Haar and LBP (Local binary patterns) training. Theory was also derived to localise the UAV within the earth-fixed coordinate system, based upon its position with respect to the landing pad within the captured image. This article briefly discusses the principles, procedures and results for landing pad detection, as well as image analysis, UAV localisation and command generation.