This work presents a new robust control scheme for articulated robotic systems. The aim of the controller is to add an adaptive scheme to the standard robust controller in order to improve its performance, especially in the case of unknown parameters or unknown variable loading. This paper presents real results for a Puma-560 robot, comparing the new strategy with the classical robust controller. They show a case in which the original controller does not exhibit a good performance, due to the presence of model uncertainties. The new strategy tries to tune automatically the uncertainty bound parameter of the robust action and results show that it has a satisfactory behaviour compared with the previous strategies.