Development of a Methodology for Dynamic Motor Control with Machine Learning

Abstract

In robotics, the problem of precision in position control of the different motor systems requires the evaluation of multiple variables in real time in order not to cause instability in the robot due to processing time. In this work, a methodology based on automatic learning of motor systems is proposed in order to be able to generate control actions in advance of their measurement and processing, for which a second-order system and its response to disturbances such as step signals are proposed.