Diseño y simulación de un controlador inteligente utilizando aprendizaje por refuerzo Q-learning para la navegación autónoma de dos robots móviles.

Trajectory planning in autonomous mobile robots is an open problem because, when working in dynamic environments, it is very expensive to program the entire navigation system for a particular application or, failing that, it would be very difficult for the programmer to correctly predict the changes...

Full description

Autores:
Carreño Puentes, Sergio Manuel
Tipo de recurso:
Trabajo de grado de pregrado
Fecha de publicación:
2022
Institución:
Universidad Antonio Nariño
Repositorio:
Repositorio UAN
Idioma:
spa
OAI Identifier:
oai:repositorio.uan.edu.co:123456789/7242
Acceso en línea:
http://repositorio.uan.edu.co/handle/123456789/7242
Palabra clave:
Aprendizaje por refuerzo,
aprendizaje automático,
aprendizaje profundo,
neurona artificial,
redes neuronales,
robots autónomos
Reinforcement learning,
machine learning,
deep learning,
artificial neuron,
neural networks,
autonomous robots.
Rights
openAccess
License
Attribution-NonCommercial-NoDerivatives 4.0 International (CC BY-NC-ND 4.0)
Description
Summary:Trajectory planning in autonomous mobile robots is an open problem because, when working in dynamic environments, it is very expensive to program the entire navigation system for a particular application or, failing that, it would be very difficult for the programmer to correctly predict the changes in the environment. In order to contribute to this field, this document shows the process of designing an intelligent controller based on reinforced learning and more specifically using the Q-learning algorithm to drive two mobile robots through a simulated environment, and that autonomously manage to learn the trajectory that will take them to a target position without having prior knowledge about the work environment