Title:
|
BUILDING A NAVIGATIONAL ENVIRONMENT FOR AUTONOMOUS AGENTS WITH REINFORCEMENT LEARNING |
Author(s):
|
Vinicius Oliverio , Cláudio Cesar De Sá , Rafael Stubs Parpinelli |
ISBN:
|
978-972-8924-97-3 |
Editors:
|
Hans Weghorn and Pedro Isaías |
Year:
|
2009 |
Edition:
|
V II, 2 |
Keywords:
|
Reinforcement learning, machine learning, artificial intelligence. |
Type:
|
Poster/Demonstration |
First Page:
|
353 |
Last Page:
|
355 |
Language:
|
English |
Cover:
|
|
Full Contents:
|
click to dowload
|
Paper Abstract:
|
The current bearing of computing could change dramatically with the understanding of how to program the computers to
learn, improve automatically with the experience. This understanding would allow the machines to do tasks that until
today they are not capable of. In this context, this work presents a computational environment directed to reinforcement
learning study for a community of autonomous mobile agents of the reactive type, where these agents learn to perform
tasks in this environment with the Q-Learning algorithm, so, these agents have the skill to learn with the experience
generated by events, feeds by a reinforcement mechanism in its learning process, using this algorithm. The studied
behaviors are the wandering, homing and following that are explained in a more detailed way along the work. After the
implementing, it's made the experimenting where the learning is validated according to the learning capacity of these
agents in different environments. |
|
|
|
|