A HUMAN AIDED LEARNING PROCESS FOR AN ARTIFICIAL IMMUNE SYSTEM BASED ROBOT CONTROL - An Implementation on an Autonomous Mobile Robot

Jan Illerhues, Nils Goerke

2007

Abstract

In this paper we introduce a pre-structured learning process that enables a teacher to implement a robot controller easily. The controller maps sensory data of a real, autonomous robot to motor values. The kind of mapping is defined by a teacher. The learning process is divided into four phases and leads the agent from a two stage supervised learning system via reinforcement learning to unsupervised autonomous learning. In the beginning, the controller starts completely without knowledge and learns the new behaviours presented from the teacher. In second phase is dedicated to improve the results from phase one. In the third phase of the learning process the teacher gives an evaluation of the states zhat the robot has reached performing the behaviours taught in phase one and two. In the fourth phase the robot gains experience by evaluating the transitions of the different behavioral states. The result of learning is stored in a rule-like association system (RLA), which is inspired from the artificial immune system approach. The experience gained throughout the whole learning process serves as knowledge base for planning actions to complete a task given by the teacher. This paper presents the learning process, its implementation, and first results.

References

  1. Hart, E., Ross, P., Webb, A., and Lawson, A. (2003). A role for immunology in 78”next generation”78 robot controllers. In ICARIS03, Artificial Immune Systems, Second International Conference, pages 46-56.
  2. KTO (1995). KTO- Kommunikation und Technologietransfer Odenthal. http://www.kurt2.de/.
  3. Rattenberger, J., Poelz, P. M., Prem, E., Hart, E., Webb, A., and Ross, P. (2004). Artifical immune networks for robot control. In Proc Fourth Int Workshop on Epigenetic Robotics: Modeling Cognitive Development in Robotic Systems, volume 117, pages 151-154.
  4. Ross, P. (2003). Notes on the RLA network implementation. Technical report, Napier University, Scotland.
  5. Sutton, R. S. and Barto, A. G. (1998). Reinforcement Learning: An Introduction. The MIT Press.
  6. Webb, A., Ross, P., Hart, E., and Lawson, A. (2003). Generating robot behaviours by evolving immune networks. Technical report, Napier University, Scotland.
Download


Paper Citation


in Harvard Style

Illerhues J. and Goerke N. (2007). A HUMAN AIDED LEARNING PROCESS FOR AN ARTIFICIAL IMMUNE SYSTEM BASED ROBOT CONTROL - An Implementation on an Autonomous Mobile Robot . In Proceedings of the Fourth International Conference on Informatics in Control, Automation and Robotics - Volume 4: ICINCO, ISBN 978-972-8865-83-2, pages 347-354. DOI: 10.5220/0001652303470354


in Bibtex Style

@conference{icinco07,
author={Jan Illerhues and Nils Goerke},
title={A HUMAN AIDED LEARNING PROCESS FOR AN ARTIFICIAL IMMUNE SYSTEM BASED ROBOT CONTROL - An Implementation on an Autonomous Mobile Robot},
booktitle={Proceedings of the Fourth International Conference on Informatics in Control, Automation and Robotics - Volume 4: ICINCO,},
year={2007},
pages={347-354},
publisher={SciTePress},
organization={INSTICC},
doi={10.5220/0001652303470354},
isbn={978-972-8865-83-2},
}


in EndNote Style

TY - CONF
JO - Proceedings of the Fourth International Conference on Informatics in Control, Automation and Robotics - Volume 4: ICINCO,
TI - A HUMAN AIDED LEARNING PROCESS FOR AN ARTIFICIAL IMMUNE SYSTEM BASED ROBOT CONTROL - An Implementation on an Autonomous Mobile Robot
SN - 978-972-8865-83-2
AU - Illerhues J.
AU - Goerke N.
PY - 2007
SP - 347
EP - 354
DO - 10.5220/0001652303470354