محمدرضا گل بهار حقیقی

Industrial robots are manipulators with high precision and repeatability making them proper alternatives to humans. In common industrial robots proportional-Integral (PI) controllers are exploited owing to their simplicity; however, they cannot guarantee appropriate and robust operation. To obtain suitable performance nonlinear controller is recommended. Sliding mode controller is a state feedback controller with fast transient response which is robust against uncertainties. Despite its advantages sliding mode controller is not a good choice for steady states due to chattering phenomenon. To solve the problem of higher order sliding mode controller and sliding mode control with boundary layer using the first method increases the computational complexity and the second method will slow down the system. On the other hand, intelligent controllers (neural networks and adaptive neuro-fuzzy network) have been successful, though they have not acceptable performance when encountering uncertainties. It is a consequence of inevitable training phase in such controllers. In this study a hybrid controller is proposed combining sliding mode and adaptive neuro-fuzzy controller with variable weights to take advantage of both structures. A fuzzy supervisor is tasked with optimal adjustment of the weights. Besides, it smooths (facilitates) switching between two controllers.