White Rose University Consortium logo
University of Leeds logo University of Sheffield logo York University logo

Towards modeling complex robot training tasks through system identification

Nehmzow, U., Akanyeti, O. and Billings, S.A. (2009) Towards modeling complex robot training tasks through system identification. Research Report. ACSE Research Report no. 993 . Automatic Control and Systems Engineering, University of Sheffield

Full text available as:
[img]
Preview
Text
993.pdf

Download (2060Kb)

Abstract

Previous research has shown that sensor-motor tasks in mobile robotics applications can be modelled automatically, using NARMAX system identi�cation, where the sensory perception of the robot is mapped to the desired motor commands using non-linear polynomial functions, resulting in a tight coupling between sensing and acting | the robot responds directly to the sensor stimuli without having internal states or memory. However, competences such as for instance sequences of actions, where actions depend on each other, require memory and thus a representation of state. In these cases a simple direct link between sensory perception and the motor commands may not be enough to accomplish the desired tasks. The contribution to knowledge of this paper is to show how fundamental, simple NARMAX models of behaviour can be used in a bootstrapping process to generate complex behaviours that were so far beyond reach. We argue that as the complexity of the task increases, it is important to estimate the current state of the robot and integrate this information into the system identification process. To achieve this we propose a novel method which relates distinctive locations in the environment to the state of the robot, using an unsupervised clustering algorithm. Once we estimate the current state of the robot accurately, we combine the state information with the perception of the robot through a bootstrapping method to generate more complex robot tasks: We obtain a polynomial model which models the complex task as a function of predefined low level sensor motor controllers and raw sensory data. The proposed method has been used to teach Scitos G5 mobile robots a number of complex tasks, such as advanced obstacle avoidance, or complex route learning.

Item Type: Monograph (Research Report)
Copyright, Publisher and Additional Information: The Department of Automatic Control and Systems Engineering research reports offer a forum for the research output of the academic staff and research students of the Department at the University of Sheffield. Papers are reviewed for quality and presentation by a departmental editor. However, the contents and opinions expressed remain the responsibility of the authors. Some papers in the series may have been subsequently published elsewhere and you are advised to cite the later published version in these instances.
Institution: The University of Sheffield
Academic Units: The University of Sheffield > Faculty of Engineering (Sheffield) > Department of Automatic Control and Systems Engineering (Sheffield) > ACSE Research Reports
Depositing User: Miss Anthea Tucker
Date Deposited: 16 Oct 2012 09:39
Last Modified: 06 Jun 2014 18:37
Status: Published
Publisher: Automatic Control and Systems Engineering, University of Sheffield
URI: http://eprints.whiterose.ac.uk/id/eprint/74646

Actions (repository staff only: login required)