Input Saturated Visual Servoing for Unmanned Aerial Vehicles

This paper presents an input saturated visual servoing controller for a quadrotor unmanned aerial vehicle (UAV). The controller regulates relative pose between the vehicle and a planar horizontal visual target. In order to simplify the control design, the method uses a virtual camera to define a set of image moment features. The image feature kinematics is independent of roll and pitch motion. Since the image feature error is stabilized in the virtual camera, the visual target potentially leaves the real camera's field of view (FoV). To keep the visual target in the camera FoV, an input saturated law is proposed to sufficiently bound the roll and pitch of the vehicle. Furthermore, an adaptive input saturated control law is proposed to ensure positive thrust. Experimental and simulation results are provided to demonstrate controller performance.