This study addresses the control problem of an unmanned quadrotor in an indoor environment where there is lack of absolute localisation data. Based on an attached inertia measurement unit, a sonar and an optic-flow sensor, the state vector is estimated using sensor fusion algorithms. A novel switching model predictive controller is designed in order to achieve precise trajectory control, under the presence of forcible wind gusts. The quadrotor’s attitude, altitude and horizontal linearised dynamics result in a set of piecewise affine models, enabling the controller to account for a larger part of the quadrotor’s flight envelope while modelling the effects of atmospheric disturbances as additive-affine terms in the system. The proposed controller algorithm accounts for the state and actuation constraints of the system. The controller is implemented on a quadrotor prototype in indoor position tracking, hovering and attitude manoeuvres experiments. The experimental results indicate the overall system’s efficiency in position/altitude/attitude set-point manoeuvres.