Fuzzy logic control of hydraulic robots with flow-deadband nonlinearities

Application of a fuzzy logic controller to a class of hydraulically actuated industrial robots is investigated in this thesis. A simple, yet effective, set of membership functions and rules are developed to meet the control requirements of such robots. The proposed PD-type fuzzy controller is first examined through simulation of a two-link hydraulic robot. The robot has the same hydraulic configuration as many existing industrial manipulators. It is shown that the controller exhibits positive aspects which cannot be easily achieved by conventional control techniques, such as a PD controller. These aspects include a short rise-time and a well maintained dampened response.; The fuzzy controller is then utilized on an instrumented Unimate MK-II robot which has been retrofitted as a research robot. An off-line routine based on the simplex method is applied to find the best performing control gains for different links. This is accomplished by minimizing the summation of errors over step input responses. The controller, although effective, is shown to produce steady-state errors. The steady-state error in a step input response is mainly due to the hydraulic valve deadband in which the control action is not effective. The steady-stat position error in a ramp input response is shown to be due to both valve deadband and the nature of the PD-type fuzzy controller.; In order to eliminate the steady-state errors, conceptually simple methods are developed and applied in parallel with the PD-type fuzzy controller. In particular, two methods are proposed. In the first method, a control term proportional to the integral of error is calculated and added to the output from the main controller. In the second method, the control measure of the steady-state error obtained from the PD-type fuzzy controller is used as an offset. Both actions are activated only in a region of fuzzy zero velocity error.