Robust control of robotic manipulators using an adaptive neural network estimator of uncertainty

This paper presents a novel robust control approach for robotic manipulators by using a neural network estimator and the voltage control strategy. The proposed estimator is an adaptive multi-layer neural network that its parameters are regulated by a propagation algorithm. The novelty of the proposed design is the use of voltage control strategy that is different with the common torque control strategy. As an advantage، it is free from the dynamical model of the robotic manipulator. Compared with the conventional robust control، it does not require the determination of the bounds of uncertainty and bounding functions. In addition، the problem of lacking data for estimation of uncertainty is solved. In fact، the back-propagation algorithm uses the tracking error and its derivative instead of the estimation error. Stability of the control system is proven by analysis. The efficiency of the proposed control approach and the estimator of uncertainty are shown by simulation of the SCARA robot driven by permanent magnet dc motors. The control performance of the proposed control is compared with a control approach which uses a fuzzy system to estimate the uncertainty. The simulation results show the superiority of the proposed control in tracking control، set point control and estimation of uncertainty.