COMPARISON OF EXTENDED JACOBIAN AND LAGRANGE MULTIPLIER BASED METHODSFOR RESOLVING KINEMATIC REDUNDANCY

Several methods have been proposed in the past for resolving the contr ol of kinematically redundant manipulators by optimizing a secondary c riterion. The extended Jacobian method constrains the gradient of this criterion to be in the null space of the Jacobian matrix, while the L agrange multiplier method represents the gradient as being in the row space. In this paper, a numerically efficient form of the Lagrange mul tiplier method is presented and is compared analytically, computationa lly, and operationally to the extended Jacobian method. This paper als o presents an improved method for tracking algorithmic singularities o ver previous work.