In the subject of quadrotor controller design, usually modelling and identification are tedious and time-consuming tasks. In this study, we propose a controller design that avoids the quadrotor's system identification procedures while staying robust with respect to endogenous and exogenous disturbances. To reach our goal, based on the cascaded structure of a quadrotor, we divide the system into positional and attitude subsystems each controlled by an independent model-free controller. We validate our control approach in two realistic scenarios : in presence of unknown measurement noise and unknown time-varying wind disturbances. We provide simulations on a realistic nonlinear quadrotor model following an aggressive position-yaw trajectory.