Visual landing insensitive to the depth with velocity and control constraints: a twisting based solution

This paper proposes a nonlinear vision based flight control law for the automatic landing of an aircraft. Although a number of methods have been proposed in this domain, our method differs by the following features. First, we do not assume knowledge of the runway size or distance from the aircraft, or the help of a ground technology to guide the aircraft. Second, this design addresses the model nonlinearities, so that it copes for initial conditions far away from the desired trajectory. Third, the control law takes input saturations into account. Fourth, it limits overshoots during the transient phase, so that the runway motion inside the videocamera image plane is limited to a minimum, and so is the risk for it to get out of the field of view. Our approach is illustrated in simulation with a simplified aircraft model. In particular good robustness properties with respect to noisy visual features are demonstrated.