Recent research into the use of kinematic redundancy, for robotic manipulators is briefly reviewed and the potential advantages of redundant manipulators over current standard manipulator designs are highlighted. Current formulations for kinematic control of manipulators are then examined with the objective of developing a generalized inverse kinematics method for redundant manipulators. It is concluded that the generalized method has to. be manipulator-independent, algorithmic and applicable for any objective furction subject to joint constraints.

Three inverse kinematics methods are developed for redundant manipulators. It is shown that a closed-form solution to the problem may be feasible and desirable from the computational time of view, however lacking generality. Attempts to achieve a complete generalized solution resulted in the development of the 'Generalized Inverse' technique for redundant manipulators based on a modified Newton iterative procedure. This method is manipulator-independent, can be applied for any objective function subject to joint constraints and be considered for real-time control.

In point-to-point motion applications, a manipulator with redundant degrees of freedom can be superior in performance to a similar standard manipulator following optimal joint trajectories. This' is accomplished through the optimization of an objective function, such as, sum of joint displacements, manipulator motion time, total energy dissipation or a combined criterion consisting of a weighted sum of several objective functions. A variety of numerical examples presented strongly support this claim.

The remainder of this study is devoted to investigate the advantages of redundant manipulators in continuous path motion applications and development of trajectory planning algorithms. 'Two approaches to continuous path planning for motion control attend effector level and joint level are developed. it is concluded that the off-line optimal ]oath planning followed by on-line trajectory control at joint level is more . advantageous for redundant manipulators. For this algorithm the desired Cartesian path is approximated by selected location nodes on the path and piecewise cubic polynomial functions are interpolated for the corresponding joint configuration nodes. The motion time of the manipulator and maximum deviation from the desired Cartesian path.are used as the objective functions. These methods provide an additional scope to the use of kinematically redundant manipulators, as illustrated in several . simulated path planning examples.