Flight and Visual Tracking Control for a Quadrotor UAV Via Image-Based Visual Servoing

Unmanned aerial vehicles are flying machines that have been used for civilian and military applications due to their properties and abilities. They combine data from accelerometers, gyros and, GPS for positioning purposes. The reliability of the navigation and location system of the aerial vehicle depends on the number of satellite signals the GPS sensor detects. These signals become weak or null if the robot is moving under certain conditions, such as cloudy skies; or in areas like closed spaces, and zones surrounded by skyscrapers or dense vegetation. Image-based visual servo-systems stand as a possible alternative for the autonomous navigation in GPS-denied environments due to its low computational cost, minimal hardware and multiple ways to complete a determined task safely. They use visual data from cameras along with information from the inertial measurement unit to estimate the position of the drone for navigation tasks. Target-tracking applications for unmanned aerial vehicles refer to those systems that make a flying robot locate over a steady object, or follow the course of a mobile objective from a desired pose. These are useful in surveillance and inspection operations, like target-approaching and landing maneuvering, vehicle chasing, and examining of power lines, structures, and wind-power generators. This dissertation introduces the design and implementation of a robust image-based visual target tracking control for a quadrotor unmanned aerial vehicle subjected to perturbations. Information regarding the position of the quadrotor is directly extracted from the image projection of the target through a virtual camera scheme and used as feedback for the dynamics of the robot. Due to its properties and finite-time convergence, an adaptive sliding mode controller is combined with the visual servoing strategy to improve its robustness against bounded disturbances and uncertainties. The adaptive gain from the sliding mode scheme grants the necessary control effort to keep the system stable, whether a perturbation exists or not. The stability of the proposal is guaranteed by a closed-loop analysis employing Lyapunov theory. Simulations and experiments validate the benefits and performance of the robust visual servo controller when tracking a static and dynamic target.