This paper presents some useful results which follow from the particular structural properties of a rigid robot model while the system is subject to the action of an output feedback. Given a rigid robot model, the controller ensures, in addition to the global asymptotic stability property, an eigenvalues assignment of the resulting linearized model within the stable region of the complex plane. In this way, required global and local control objectives can be achieved. Furthermore, the design of the controller is accomplished by applying a sort of a decoupling procedure that decomposes the entire nonlinear closed-loop system to a set of reduced-order nonlinear systems. The dependence of the eigenvalues of the linearized model on the model uncertainties is investigated. Simulation results that demonstrate the potential of the approach are presented.