Neural control of a Quadrotor: A state-observer based approach

Abstract

This paper proposes a neuro-adaptive controller with supervisory control for UAVs. A state observer of the control errors is defined because the vehicle could be affected by measurement noise and external disturbances. In this study, the control errors are estimated by a state neural observer from which the neural adaptive system generates a control action. Also, a supervisor term is defined in order to guarantee the system stability. The neuro-adaptive controller parameters are updated on-line by the adjustment laws that are a function of the control errors calculated by the observer. These adjustment laws were obtained through the Lyapunov stability criterion. To summarize, simulation results demonstrate the good performance of the proposed technique for trajectory tracking tasks in navigation environments affected by measuring noise and external disturbances.