Development of a Quadrotor with Vision-based Target Detection for Autonomous Landing

Abstract

In the field of robotics, the quadrotors have rapidly gained interest and have made several breakthroughs involving it, which range from variable pitch to application of swarm robotics. With that said, this paper aims to also expand upon one of the current developments which is the automated landing of quadrotors on a designated landing zone. Without GPS, the prototype built in this research employs image processing techniques to detect the landing zone, as well as to determine the flight altitude. Using these information, the quadrotor is autonomously controlled via its control surfaces (throttle, roll and pitch) in order to perform the landing procedure. Additionally, the quadrotor is capable of tracking a moving target and safely land even with winds reaching up to 2.2m/s.