Nonlinear Model Predictive Control for Robot Manipulator with Disturbance Observer

This paper investigates a variable horizon nonlinear model predictive control strategy based on disturbance observer for a three-degree-of- freedom robot manipulator which has the property of high nonlinearity and uncertainty. Firstly, a nonlinear disturbance observer is implemented to accurately estimate and compensate for uncertain factors such as external disturbances, model parameter disturbance and unmodeled part of the system. Secondly, based on the disturbance observer, the nonlinear model predictive control is utilized to handle the nonlinear model such that avoiding large model error and the predictive horizon of nonlinear model predictive control is improved to the form of dynamic optimization, and the optimal predictive horizon is selected according to the state error of each sampling period. Compared with constant predictive horizon nonlinear model predictive control, this improved strategy not only improves the system tracking performance, but also reduces the use of computing resources and shortens the calculation time. Finally, the numerical simulation and comparative experimental results are presented to verify the superiority and effectiveness of the proposed control strategy.