Image-Based Visual Servo Control for a Quadrotor UAV

Abstract

This dissertation presents an adaptive image-based visual servoing (IBVS) integrated with adaptive sliding mode control for a vision-based operation of a quadrotor unmanned aerial vehicle (UAV). For a seamless integration with under-actuated quadrotor dynamics, roll and pitch channels are decoupled from the other channels using virtual features. This allows a simple and accurate algorithm for estimating depth information and successful application of the proposed guidance and control algorithm. By employing an adaptive gain in the IBVS control method, the chance of image feature loss is reduced, and performance and stability of the vision-guided UAV control system are improved. The overall setup allows image features to be placed at the desired position in the image plane of a camera mounted on the quadrotor UAV. Stability of the IBVS system with the nonlinear adaptive controller is proved using Lyapunov stability analysis. Performance of the overall approach is validated by numerical simulation, vision-integrated hardware-in-the-loop simulation and experiments. The results con_x000C_rm that the target image is successfully placed at the desired position of the image plane, and the quadrotor state variables are properly regulated, showing robustness in the presence of sensor noise, parametric uncertainty, and vibration from motors. For further applications of the proposed IBVS-guided control algorithm, target tracking and autonomous landing on a moving platform are demonstrated.