In this project, an intelligent controller was proposed for high-accuracy position tracking control of a 6-degree-of-freedom (6DOF) robot. The control scheme is formatted using a 2-layer neural-network template, in which the first layer is encoded by a proportional-derivative structure, and the second layer is comprised of three different functions. Three ranges of perturbation that affects to the control performance are respectively treated by an offset learning, a folding PD control and high-switching control terms. The gains of the control terms are updated using modified gradient methods. Effectiveness of the proposed control is successfully verified by comparative simulation results and real-time experiments on a Unimation 6DOF robot.
Authors: Vo Tan Tai, Doan Ngoc Minh, and D. X. Ba