Synthesis of a Robust Neurocontroller in the Face of Strong External Disturbances

This study focuses on the synthesis of a robust neurocontroller for a direct drive robotic manipulator having two degrees of freedom. The method presented in the paper uses the Variable Structure Systems methodology in stabilizing the gradient descent based training dynamics of the controller. The use of this approach introduces the robustness of variable structure systems based design framework into the design, which is subject to the varying environmental conditions such as payload changes and noisy observations. Simulations have been carried out with the conditions stated above and the results observed clearly recommend the use of the algorithm presented.