Control visual de un vehiculo aereo autonomo usando deteccion y seguimiento de caracterısticas en espacios exteriores

Abstract

The main goal of this thesis is to build vision-guided autonomous flying robots. Tasks like feature detection, target tracking, obstacle avoidance using vision sensors allows such robots to serve as intelligent eyes-in-the-sky suitable for numerous applications including law enforcement, search and rescue, aerial mapping and inspection, and movie making. Furthermore, computer vision may reduce uncertainty and increase versatility and overall accuracy of robotic tasks which are important concerns in most applications. We address the visual servoing problem for an Unmanned Aerial Vehicle (UAV) in outdoor environments, specifically an autonomous robotic helicopter. We propose vision-based techniques which allow an UAV to perform maneuvers towards features of interest when GPS has dropouts (usually in urban areas) or to track a target. We investigate visual servo control techniques that use velocities of suitable image features parameters directly to compute the references for the flight control for driving the robot, i.e, the tasks is specified directly in the sensor workspace. Therefore, the strategy does not require camera calibration procedures or 3D scene reconstruction schemes which are subject to errors and demand high processing power. Although visual servoing is a well studied problem for ground-based robots or robotics manipulators operating in 2D environments, is not well studied for the visual control problem of an UAV such as the one proposed in this thesis. The approach proposed here was analyzed and validated using several experimental tests on different platforms. We investigate a number of factors that influence the performance of the system including processing frame rate, vibrations, control strategies and environmental conditions such as light and luminance, background changes, etc. Experiments on the real autonomous helicopter show that visual servoing approach can be used to control the displacements of an autonomous helicopter vertically, laterally and longitudinally, and can be used to both, track an external target and guide the UAV trajectory.