Diseño, programación y simulación de controladores inteligentes aplicados a un robot móvil de tracción diferencial: diseño, programación y simulación de un controlador basado en redes neuronales aplicado a un robot móvil de tracción diferencial para seguimiento de trayectorias.

Abstract

This document presents an analysis for the implementation of intelligent controllers based on neural networks for trajectory tracking of a mobile robot. The system is simulated using the Virtual Robot Experimentation Platform (V-REP) software, which contains an integrated development environment based on a distributed control architecture with objects that can be controlled through a script, created within the platform itself. or commanded by a remote client Application Programming Interface (API). The Pioneer 3XD differential mobile robot will be used and through the remote client platform, the robot will be commanded from Matlab, since this software has a perfected desktop environment for the analysis and numerical calculation of complex algorithms such as intelligent controllers ( for example, based on neural networks) to be developed in this project. The proposed scheme consists of a cascade design, the external loop allows a kinematic controller based on the inverse model for position control that allows trajectory tracking and an internal loop with dynamic controllers based on neural networks that, when faced with changes in the environment of the robot (inclined plane), allows to control its linear and angular speed. Controllers based on neural networks are created from an uncertain model, which will be tested in different types of trajectories: square, circular, lemniscate and open. Finally, a comparative analysis between the intelligent controllers and the conventional PID controller will be shown to show the performance qualitatively and quantumly through ISE, IAE and ISCO performance indices.