DECENTRALIZED ADAPTIVE-CONTROL OF MANIPULATORS

This article presents two new adaptive schemes for motion control of r obot manipulators. The first controller possesses a partially decentra lized structure in which the control input for each task variable is c omputed based on information concerning only that variable and on two ''scaling factors'' that depend on the other task variables. The need for these scaling factors is eliminated in the second controller by ex ploiting the underlying topology of the robot configuration space, and this refinement permits the development of a completely decentralized adaptive control strategy. The proposed controllers are computational ly efficient, do not require knowledge of either the mathematical mode l or the parameter values of the robot dynamics, and are shown to be g lobally stable in the presence of bounded disturbances. Furthermore, t he control strategies are general and can be implemented for either po sition regulation or trajectory tracking in joint-space or task-space. Computer simulation results are given for a PUMA 762 manipulator, and these demonstrate that accurate and robust trajectory tracking is ach ievable using the proposed controllers. Experimental results are prese nted for a PUMA 560 manipulator and confirm that the proposed schemes provide simple and effective real-time controllers for accomplishing h igh-performance trajectory tracking. (C) 1994 John Wiley & Sons, Inc.