Robot Axis Control Using a Differential Learning Algorithm

B. Abegaz
Loyola University of Chicago,
United States

Keywords: robots, control, unsupervised-learning, machine-learning, motors


The motion control of robotic arms using brushless DC motors has been implemented using various mechanisms, including a model predictive controller (MPC), a fuzzy controller and supervised machine learning methods. In order to provide a more efficient means of navigation, a new type of unsupervised learning mechanism based on a novel differential clustering algorithm has been designed. The differential clustering control method is useful for real time motion planning of robots and autonomous systems, where standard geometric and topological models may not work as desired due to the large operational complexity of such autonomous systems. The control mechanism has been implemented for a six degrees of freedom robotic arm and the axis control results were compared with the results of implementing MPC, fuzzy, neural network and a supervised classification based machine learning control approach. It was observed that the clustering based robotic arm control method provides comparably superior control of the torque (Nm), the speed (rpm), the angle (rad) and the angular speed (rad/s) of the robotic arm.