Fuzzy learning grasping force controller for manipulator hand

A fuzzy set adaptive and learning control algorithm was developed. It was implemented with a digital signal processor. In order to confirm the feasibility of applying it to practical systems, the digital controller was experimentally applied to grasping force control for a manipulator hand powered by a transistor PWM converter-fed servomotor. This robust manipulator hand can grip objects with different compliances stably. The controller is applicable to systems whose dynamics change time-variantly. Usually adaptive control cannot be applied to such systems since it takes time for an adaptive mechanism to converge to the final state. The proposed learning algorithm is derived on the basis of sliding mode control. It gives strong error convergence properties, which are proved by using the Lyapunov stability theorem, since it does not contain integrators. It can be applied to adaptive algorithms for controller and filters and to a learning algorithm for a neural network used in a control system