MPC for Robot Manipulators With Integral Sliding Modes Generation

This paper deals with the design of a robust hierarchical multiloop control scheme to solve motion control problems for robot manipulators. The key elements of the proposed control approach are the inverse dynamics-based feedback linearized robotic multi-input-multi-output (MIMO) system and the combination of a model predictive control (MPC) module with an integral sliding mode (ISM) controller. The ISM internal control loop has the role to compensate the matched uncertainties due to unmodeled dynamics, which are not rejected by the inverse dynamics approach. The external loop is closed relying on the MPC, which guarantees an optimal evolution of the controlled system while fulfiling state and input constraints. The motivation for using ISM, apart from its property of providing robustness to the scheme with respect to a wide class of uncertainties, is also given by its capability of enforcing sliding modes of the controlled system since the initial time instant, allowing one to solve the MPC optimization problem relying on a set of linearized decoupled single-input-single-output (SISO) systems that are not affected by uncertain terms. The proposal has been verified and validated in simulation, relying on a model of a COMAU Smart3-S2 industrial robot manipulator, identified on the basis of real data.