Control Neuronal en Línea para Regulación y Seguimiento de Trayectorias de Posición para un Quadrotor

Automatic control systems every day become more important in everyday life; therefore, it must find new and better ways to incorporate mathematical models and adaptive control algorithms to cope with a number of technical and physical challenges for exploitation. In this paper, the algorithm of the dynamic model of a Quadrotor applied to an angular position and trajectory control as a study case is detailed. Due to nonlinear nature of this type of systems, an adaptive on line neurocontroller algorithm based on B-spline neural networks is proposed, the learning procedure is divided in two stages: a) an initial off line training and; b) an on line continuous learning. This form of learning allows the Quadrotor extend its satisfactory performance at different operating conditions and trajectory tracking. The simulation results demonstrate the applicability of the developed model and the impact of dynamic control on the system performance, diminishing the exact model requirement and the possibility to incorporate the system non linearities.