A Proposal of a Methodology for the Analysis of Robot-Environment Interaction through System Identification

Otar Akanyeti, Iñaki Rañó, Ulrich Nehmzow, S. A. Billings

Research output: Chapter in Book/Report/Conference proceedingConference Proceeding (Non-Journal item)

8 Downloads (Pure)

Abstract

This paper presents a novel method to analyse robot-environment interaction from mathematical point of view in order to understand and identify the underlying rules governing this interaction. The method finds its theoretical roots in robot training, system identification and dynamical systems theory.
The method is based on identifying the while system using two models – i) the controller model which describes how the actions of the robot are related to its own perception and ii) the perception model which describes how the perception of the robot changes as a result of the actions of the robot – and then put them together into a single coupled dynamical system in order to analyse or simulate the emergent behaviour.
We tested the proposed methodology by analysing the robot-environment interaction where the desired robot behaviours was a door traversal task.
Original languageEnglish
Title of host publicationProceedings of Towards Autonomous Robotic Systems 2008
EditorsSubramanian Ramamoorthy, Gillian M. Hayes
PublisherEdinburgh University Press
Pages43-49
Number of pages7
ISBN (Electronic)9781906849016
ISBN (Print)9781906849009
Publication statusPublished - 30 Sept 2008
Externally publishedYes

Fingerprint

Dive into the research topics of 'A Proposal of a Methodology for the Analysis of Robot-Environment Interaction through System Identification'. Together they form a unique fingerprint.

Cite this