Seguimiento de objetos basado en visión artificial para cuadrirrotor Parrot AR.Drone 2.0

Abstract

The purpose of this project is to develop, implement and evaluate a system capable of performing the autonomous tracking of an object using the Parrot AR.Drone 2.0 airborne robot. To achieve this, both artificial vision and automatic control techniques were used. This implementation made it possible to detect the selected objects of interest, as well as, allowed the tracking of moving objects in a robust way using the Drone. The project presents an implementation to follow an object in three degrees of freedom under a controlled environment. We used the ros_opentld computational package under the ROS Jade Framework in Ubuntu 14.04, which provides the position data of the object with respect to the front camera of the Parrot AR.Drone 2.0. The data acquired from the object image, using ros_opentld, were used for the development of three PID controllers. Each of them allowed to control the flight of the Drone in each one of its three axes. In addition, state estimators and low pass filters are developed to improve the efficiency and stability of the system.