“…The core of the scheme consists in the feedforward compensation of the inverse dynamics. Such type of compensation is preferable then the feedback one, since the latter requires the measurement or at least the precise knowledge of the endeffector's pose, translational and angular velocities, which is not easy to manage without additional and expensive sensors (Abdellatif and Heimann, 2007;Abdellatif et al, 2005;Burdet et al, 2000;Kim et al, 2005;Wang and Ghorbel, 2006). The feedback controllers of the single actuators are kept linear and simple to avoid additional computational effort.…”