Offset-free nonlinear model predictive control for improving dynamics of cable-driven parallel robots with on-board thrusters

Abstract

In this article, thrusters embedded on a cable-driven parallel robot (CDPR) platform are proposed to improve the CDPR dynamics and trajectory tracking performance. On-board thrusters with their short response time can compensate for the reduced bandwidth of the winch actuation due to winding speed limit and low cable stiffness. To compute and allocate control signals to winches and thrusters, a nonlinear model predictive controller (NMPC) is designed. A model of a CDPR with its hybrid actuation dynamics and saturation is introduced, including an alternative model of the thrust actuation that removes the need of thrust sensors and reduces the size of the NMPC optimization problem. To achieve a zero steady-state error with the NMPC, which becomes offset-free, an augmented model including additional disturbance states is proposed. The theoretical conditions to achieve offset-free control are checked. The proposed NMPC scheme is validated experimentally on a planar CDPR with three cables and four propeller-based thrusters. Results show that onboard thrusters contribute to a tracking error reduction along complex trajectories and an efficient damping of the platform vibration.