Quadrotor Visual Servoing for Automatic Landing

This thesis investigates the application of camera-based tracking and landing for multirotor UAVs on a moving target. Both Position-Based and Image-Based Visual Servoing are utilized to provide guidance to the quadrotor which will aim to track the target and to land on top of it. A novel Image-Based Visual Servoing scheme with a non-static desired image is developed to facilitate landing and yawing motions by the quadrotor. These control laws are then investigated through simulation and implementation on an experimental platform. Experiments are performed using the Ardrone 2.0 quadrotor and a Create Roomba as the ground rover. Successful implementations for landing are presented in the case of the Position-Based Visual Servoing, while hover and small motions are performed using the Image-Based Visual Servoing.M.A.S