Robust visual servoing using adaptive sliding mode control and quaternions for a quadrotor UAV tracking a dynamic target

Abstract

Quadrotor unmanned aircraft vehicles are being used nowadays more than ever in dangerous and complicated environments that include on-sea structure inspections, density-forest analy- sis, search-rescue operations, and precision agriculture activities, among many others. Given its extended use in multiple scenarios, the need for proper control and robustness against complex environments is highlighted, resulting in a reluctant adoption of quadrotors for fully autonomous operations. This is especially true in applications where local positioning is re- quired such as target tracking operations, object picking, indoor navigation, between others. Most of the time these operations are executed using an onboard camera, as such, keeping the objective inside the eld of view is a major problem when external perturbations affect the aircraft. To propose a solution to this problem, this thesis presents a robust image-based visual servoing -control- design for a quadrotor unmanned aerial vehicle performing visual target-tracking operations in the presence of turbulent winds. Visual data, extracted by the analysis of critical image features, is processed to control the positioning and heading of the aerial vehicle. The image acquisition algorithm considers a virtual camera approach, which produces an image insensitive to the roll and pitch movements. The previous image operations and the quadrotor modeling are performed using the quaternion rotational repre- sentation, which avoids many of the well-known Euler angle singularities. Additionally, a novel adaptive non-singular fast terminal sliding mode strategy is introduced to minimize the visual servoing error. Unlike other sliding mode methods, the proposed approach reduces the complexity of the system due to the reduction of control parameters, while providing practical nite-time convergence, robustness against bounded external disturbances as well as model uncertainties, non-overestimation of the control gains, and chattering attenuation. Furthermore, the stability of the system in a closed loop is guaranteed through the Lyapunov stability analysis. Finally, the proposed control algorithm is extensively tested using the Gazebo/ROS simulator which provides a close-to-real-life insight into the performance of the system comparing it to the same scenario using the Euler angles representation.