This paper presents an adaptive scheme for the motion control of kinem
atically redundant manipulators. The proposed controller is very gener
al and computationally efficient since it does not require knowledge o
f either the mathematical model or the parameter values of the robot d
ynamics, and is implemented without calculation of the robot inverse d
ynamics or inverse kinematic transformation. It is shown that the cont
rol strategy is globally stable in the presence of bounded disturbance
s, and that in the absence of disturbances the size of the residual tr
acking errors can be made arbitrarily small. The performance of the co
ntroller is illustrated through computer simulations with a nine degre
e-of-freedom (DOF) compound manipulator consisting of a relatively sma
ll, fast six-DOF manipulator mounted on a large three-DOF positioning
device. These simulations demonstrate that the proposed scheme provide
s accurate and robust trajectory tracking and, moreover, permits the a
vailable redundancy to be utilized so that a high bandwidth response c
an be achieved over a large workspace.