Controle preditivo aplicado ao manipulador DENSO-VP6242 utilizando quatérnios duais

Abstract

This work presents an application of non-linear model predictive control (NMPC), combined to a kinematic model based in dual quaternion of a comercial robotic manipulator. The 6—DOF serial manipulator, VP-6242 by Denso Robotics was used as reference. Having developed the system’s explicit model, optimal control actions are computed in a prediction horizon according to an objective function to be minimized. The optimization problem, considers the structure constraints determined by the robotic system, guaranteeing physical integrity. Dual quaternion algebra adds robustness to the model regarding singularities and computational advantages when compared to traditional homogeneous transformation matrices methods. For NMPC, is possible to deal with multiple variable problems of non-linear nature, which solution is provided by MATLAB MPC Toolbox. Simulation results are presented, showing that the implemented algorithm is capable of generating safe and smooth routes to the highly complex system.