Wind gust estimation for precise quasi-hovering control of quadrotor aircraft

This paper focuses on the control of quadrotor vehicles without wind sensors that are required to accurately track low-speed trajectories in the presence of moderate yet unknown wind gusts. By modeling the wind disturbance as exogenous inputs, and assuming that compensation of its effects can be achieved through quasi-static vehicle motions, this paper proposes an innovative estimation and control scheme comprising a linear dynamic filter for the estimation of such unknown inputs and requiring only position and attitude information. The filter is built upon results from Unknown Input Observer theory and allows estimation of wind and vehicle state without measurement of the wind itself. A simple feedback control law can be used to compensate for the offset position error induced by the disturbance. The proposed filter is independent of the recovery control scheme used to nullify the tracking error, as long as the corresponding applied rotor speeds are available. The solution is first checked in simulation environment by using the Robot Operating System middleware and the Gazebo simulator and then experimentally validated with a quadcopter system flying with real wind sources.