Challenges and solutions for sensor-based incremental nonlinear control of hydraulically actuated parallel robots

Abstract

Incremental Nonlinear Dynamic Inversion has shown increases in performance and robustness to model mismatches and uncertainties in hydraulic control as compared to other model-based controllers. This work will expand on hydraulic Incremental Nonlinear Dynamic Inversion force control and discuss three of its challenges and accompanying solutions. The solutions are tested experimentally on the SIMONA Research Simulator. First, the importance of the synchronization of the linearization loops of the Incremental Nonlinear Dynamic Inversion controller is shown analytically and through experiments. Secondly, it is found that saturation of the electro-hydraulic servo-valve leads to wind-up when integral action is present. Pseudo Control Hedging is implemented to deal with the wind-up effects due to saturation. The implementation of the Pseudo Control Hedging is evaluated through experiments on the SIMONA Research Simulator. Thirdly, it is shown analytically that the main spool measurements should be used for the control increment of the Incremental Nonlinear Dynamic Inversion controller. Measurements of the main spool position are therefore used in the control, but it is often difficult to extract these from the servo-valve. Through experiments it is found that measurements of the main spool position can be replaced by either a first- or second-order servo-valve model.