Direct adaptive control of electrically driven robot manipulators using quantum neural networks as uncertainty estimator

This paper develops an adaptive controller for electrically driven robot manipulators using quantum neural networks (QNN). Recently, some controllers for nonlinear and chaotic systems have been proposed using QNN. However, the tuning process of QNN is performed using offline optimization algorithms or the gradient descent algorithm. The contribution of this paper is introducing a direct adaptive controller based on QNN using Lyapunov stability theorem. In fact, the parameters of QNN are estimated by applying the adaptive rules obtained by Lyapunov theorem. Due to the nonlinear relation between the output of QNN and its tuning parameters, Taylor series expansion is needed to obtain a linear relation which is ideal for adaptive control. Simulation results on a robotic system show that the control action is successful with small tracking error. Also, in comparison with radial basis function (RBF) neural networks, QNN is more powerful in disturbance rejection.