Adaptive tracking control of robotic manipulator with uncertain dynamics based on a new neural approximator

In controlling robotic manipulator, compensating for uncertain dynamics based on the well-known neural network (NN s) approximator is a long -standing, yet well documented open problem in the field of adaptive control. In fact, the existing research and simulation examples have illustrated the effectiveness of the NN s adaptive controller on the manipulator system. However, the online update of the entire NN s idealized weight matrix (IWM) will lead to burdensome computation to the system as the NN s nodes increase. So far, there is still no suitable solution to this problem. In this paper, we propose a novel adaptive scheme based on optimized NNs approximating mechanism. Technically, by using the NN s to compensate the uncertain dynamics and unknown torques online, and by estimating and updating the square of the maximum singular value of the IWM, the burdensome online computation is successfully avoided. Furthermore, it is shown that the asymptotic convergence of the tracking error and the NN s parameters estimation error to arbitrarily small neighborhood of the origin can be guaranteed simultaneously. The illustrative examples shows the effectiveness and superiority of the proposed control scheme.