Fuzzy PID tuning for robot manipulators

The performance of any PID controller depends on the specific parameters of the system. Once these parameters change, a significant amount of effort may be required to manually tune the gains. Only a few conventionally designed robots are able of coping with changes in robot parameters. In such systems, gains are usually kept very high in order to suppress the effects of change of load and disturbances. The solution to the problem lies in the development of novel control methods that control the overall movements of the robot whilst simultaneously correcting the errors. We present a simple fuzzy tuning procedure for the PID gains K/sub P/, K/sub I/ and K/sub D/. The new procedure varies the PID gains according to the momentary robot configuration and assumes large gain values only when really needed. The effectiveness of the proposed controller is demonstrated through computer simulations for a robot manipulator with two degrees of freedom.