Robust neural network‐based backstepping landing control of quadrotor on moving platform with stochastic noise

Abstract

AbstractThis article introduces a novel nonlinear control approach for quadrotor landing on a moving ship. The dynamics of the relative position and attitude dynamics with uncertainties, unmodeled dynamics, external disturbances, and control input saturation is derived. Then, a novel robust adaptive constrained backstepping control law for the propeller and torques of the quadrotor is designed. To deal with uncertainties and un‐modeled dynamics, radial basis function is employed. The bounded‐in‐probability stability is used to alleviate the destructive influence of the stochastic Wiener process on the system positions and attitudes. Whereas the quadrotor is under‐actuated, a redundant control input technique, which affects the attitude commands, is presented. The actual control input saturation limits are turned into unknown saturation limits of the redundant control inputs. Also, since two phases for the landing problem are considered, a smooth switching mechanism between the control inputs is developed. Simulation results show the merits and applicability of the developed robust adaptive nonlinear controller.