Motion Control of 3 Degree-Of-Freedom Direct-Drive Robot

Abstract

Modern motion controllers of robot manipulators require knowledge of the system's dynamics in order to intelligently predict the torque command. The main objective for this thesis is to apply various motion controllers on a parallel direct drive robot in simulations and verify if one can take advantage of the model knowledge to improve performance of controllers. The controllers used in this thesis varied from simple PD control with position and velocity reference only applied independently at each joint to more advanced PD control with full dynamic feedforward term and computed torque control, which incorporate full dynamic knowledge of the manipulator. In the first part, a thorough study of deriving dynamic equation using Lagrange formulation has been presented as well as the actual derivation of dynamic equations for MINGUS2000. Next, in order to prepare proper sets of inputs for the simulations, detailed discussions of end effector trajectory path planning and inverse kinematics determination have been presented. Finally, background theories of various controllers used in this thesis have been presented and their simulation results on the closed-chain direct drive robot have been compared for verification purposes.