A practical Visual Servo Control for a Unmanned Aerial Vehicle

Abstract

An image-based visual servo control is presented for an unmanned aerial vehicle (UAV) capable of stationary or quasi-stationary flight. The proposed control design addresses visual servo of 'eye-in-hand' type systems. The control of the position and orientation dynamics are decoupled using a visual error based on a spherical centroid data, along with estimation of the gravitational inertial direction. The error used compensates for the poor conditioning of the Jacobian matrix seen in earlier work in this area by introducing a non-homogeneous gain term adapted to the visual sensitivity of the error measurements. A nonlinear controller is derived for the full dynamics of the system. Experimental results on an experimental UAV known as an X4-flyer made by the French Atomic Energy Commission (CEA) demonstrate the robustness and performances of the proposed control strategy.