Fuzzy inference and the control of flexible robotic manipulators

The authors discuss the implementation of controllers for robotic manipulators with flexible links by using a class of fuzzy inference systems. A summary of basic concepts which describe fuzzy logic controllers as a class of fuzzy inference systems is presented. It is shown that fuzzy inference systems can be regarded as linear models, i.e. they can be represented by a linear combination of a set of basis functions. This type of system is structurally similar to those in the class of neural networks consisting of a collection of locally-tuned units; therefore, coming up with the appropriate set of fuzzy rules to generate the corresponding fuzzy basis function expansion is analogous to the problem of efficiently placing the centers of the receptive fields of units in the neural network case. An efficient orthogonal least squares algorithm is used to accomplish this task. The technique is applied to the implementation of a simple controller to control a two link manipulator consisting of one rigid and one flexible links