Model predictive control (MPC) of robot arm

Multi joint robot arms are frequently used in unmanned autonomous vehicle applications and in many areas of the industry. The theoretical models parameter complexity obtained from the dynamic and kinematic equations of the robot arms increases with the number of joints. For high-precision applications, the model parameter complexity in model is important in the control of the robot arm. In addition to that linkage friction, internal disturbance, and external disturbance effect have been important research topics in the effective control of the robotic arm. In this thesis study, the theoretical models of robot arms which have RR (Revolute Revolute), RRR (Revolute Revolute Revolute) and RPR (Revolute Prismatic Revolute) joints structure are constructed in vector-matrix form to include all parameters. It has been determined that the M(q) mass, V(q,q ̇) coriolis-centrifugal and G(q) gravity equations obtained in the matrix form are nonlinear. The slip at the operating point of the robot arm is determined when the external disturbance effect was not well defined or when there are variable external disturbance effects in the environment. In order to control the robustness of the multi-jointed robot arm under diffrent variable external disturbances, model-based control is carried out instead of a conventional control. The model of multi-joint robot arm is obtained by not only linear but also nonlinear system identification methods and the model parameters are estimated. Model Predictive Control (MPC) of robot arm with RR and RRR joint structure is performed with the obtained model parameters. The developed algorithm for the robotic arm model obtained under the external disturbance was embedded on the RRR jointed robot arm (gimbal) on the Unmanned Aerial Vehicle (UAV) and the MPC control of the gimbal was performed. The experimental results are compared with the results from the conventional control methods and the performance of the proposed algorithm has been tested under different conditions. It seems that the MPC has succeeded in performance criteria such as speed, robustness, sensitivity and allowing constraints.