An approach to resolve the kinematic redundancy and to control the motion of a redundant manipulator is described. The redundancy is resolved by imposing additional constraints on the arm’s motion. These constraints may arise, for example, from physical restrictions or from necessary conditions of an optimization problem. The formulation proposed leads to a reduction in the number of independent system coordinates and thus to a reduced order dynamical model. Using this model, a control architecture is designed so that the task space variables and the forces of constraints are decoupled, which facilitates the control design. A simulation example is described to illustrate the method.