A stiffness control of a manipulator using a fuzzy model

In this paper, we suggest a method of deciding the PD gains of a stiffness controller using an identification method based on the Takagi-Sugeno´s fuzzy model. It is difficult to perform a compliance task due to the characteristics of the robot itself and an uncertain work environment. Therefore, in this paper, we identify a fuzzy model by dividing the relationship of input-output data into several piecewise-linear equations using the Hough transform, which is an image processing method, and by fine-tuning the parameters of the fuzzy model using the gradient-descent method. By using this fuzzy model, we propose a method of designing the PD gains of the stiffness controller. Finally we show the validity of this method by a surface tracking experiment using a paper box