A robust MPC/ISM hierarchical multi-loop control scheme for robot manipulators

In this paper, we propose a robust hierarchical multi-loop control scheme aimed at solving motion control problems for robot manipulators. The kernel of the proposed control scheme is the inverse dynamics-based feedback linearized robotic MIMO system. A first loop is closed relying on an Integral Sliding Mode (ISM) controller, so that matched disturbances and uncertain terms due to unmodelled dynamics, which are not rejected by the inverse dynamics approach, are suitably compensated. An external loop based on Model Predictive Control (MPC) optimizes the evolution of the controlled system in the respect of state and input constraints. The motivation for using ISM, apart from its property of providing robustness to the scheme in front of a significant class of uncertainties, is also given by its capability of enforcing sliding modes of the controlled system since the initial time instant, which is a clear advantage in the considered case, allowing one to solve the model predictive control optimization problem relying on a set of linearized decoupled SISO systems which are not affected by uncertain terms. As a consequence, a standard MPC can be used and the resulting control scheme is characterized by a low computational load with respect to conventional nonlinear robust solutions. The verification and the validation of our proposal have been carried out with satisfactory results in simulation, relying on a model of an industrial robot manipulator with injected noise, to better emulate a realistic set up. Both the model and the noise have been identified on the basis of real data.