Diseño de un modelo para la detección de obstáculos mediante la utilización de las redes neuronales artificiales

Abstract

The present work intends to take advantage of the generalization capacity of neural networks to solve the problems that arise in the navigation of a mobile robot in an unknown environment. Today, science and technological advances have been able to develop vertically and objectively neural networks through a number of tools for their design, vast have basic knowledge of the operation of them, one of the most used is Matlab Which has a series of bookstores dedicated exclusively to this purpose, in addition provides a friendly environment for the programmer. This thesis is divided in 4 phases: In the first one the objective of the investigation is presented that includes the study of the characteristics of the neural networks, recognition of patterns and robots; The second describes the design to be developed in the virtual environment SimRobot; In the third phase the algorithms developed for the model are applied together with the experimental tests described in this phase, the fourth phase is about the discussions of the research project and in the last one the conclusions and recommendations are presented in the future .