Model Predictive Control of 2-Degree of Freedom Robotic Manipulator with Backlash

Abstract

This thesis presents the design of model predictive control of 2-degree of freedom robotic manipulator with backlash at the input of a stable linearized system under control constraints. This system is a two-axis serial-robotic manipulator with a nonlinear dynamic model and due to its non-linear nature, it is very useful in teaching environments and research. Backlash is a very important non-linearity that limits the performance of speed and position control in industrial, and in many other applications. The control objective is output tracking and disturbance rejection, we have formulated a multi-objective optimization problem with a cost/objective function having terms representing closed-loop performance. Our main goal is to track output in the presence of nonlinearities. The resulting optimal control problem is implemented online via a Model Predictive Control (MPC) scheme. In this context, MPC arises a powerful technique for achieving multiple objectives for the MIMO system. The reason for the use of MPC is that it can easily handle the multivariable case, system constraint, and non-linearities in a very simple way.