Learning object relationships which determine the outcome of actions

Severin Fichtl, John Alexander, Dirk Kraft, Jimmy Alison Jorgensen, Norbert Kruger, Frank Guerin

Research output: Contribution to journalArticle

4 Downloads (Pure)

Abstract

Infants extend their repertoire of behaviours from initially simple behaviours with single objects to complex behaviours dealing with spatial relationships among objects. We are interested in the mechanisms underlying this development in order to achieve similar development in artificial systems. One mechanism is sensorimotor differentiation, which allows one behaviour to become altered in order to achieve a different result; the old behaviour is not forgotten, so differentiation increases the number of available behaviours. Differentiation requires the learning of both sensory abstractions and motor programs for the new behaviour; here we focus only on the sensory aspect: learning to recognise situations in which the new behaviour succeeds. We experimented with learning these situations in a realistic physical simulation of a robotic manipulator interacting with various objects, where the sensor space includes the robot arm position data and a Kinect-based vision system. The mechanism for learning sensory abstractions for a new behaviour is a component in the larger enterprise of building systems which emulate the mechanisms of infant development.
Original languageEnglish
Pages (from-to)188-199
Number of pages12
JournalPaladyn
Volume3
Issue number4
DOIs
Publication statusPublished - 1 Dec 2012

Fingerprint

Manipulators
Robotics
Learning
Robots
Sensors
Industry
Object Attachment
Child Development

Keywords

  • Developmental Artificial Intelligence
  • vision
  • infant development
  • means-end behaviour
  • learning preconditions

Cite this

Fichtl, S., Alexander, J., Kraft, D., Jorgensen, J. A., Kruger, N., & Guerin, F. (2012). Learning object relationships which determine the outcome of actions. Paladyn , 3(4), 188-199. https://doi.org/10.2478/s13230-013-0104-x

Learning object relationships which determine the outcome of actions. / Fichtl, Severin; Alexander, John; Kraft, Dirk; Jorgensen, Jimmy Alison; Kruger, Norbert; Guerin, Frank.

In: Paladyn , Vol. 3, No. 4, 01.12.2012, p. 188-199.

Research output: Contribution to journalArticle

Fichtl, S, Alexander, J, Kraft, D, Jorgensen, JA, Kruger, N & Guerin, F 2012, 'Learning object relationships which determine the outcome of actions' Paladyn , vol. 3, no. 4, pp. 188-199. https://doi.org/10.2478/s13230-013-0104-x
Fichtl S, Alexander J, Kraft D, Jorgensen JA, Kruger N, Guerin F. Learning object relationships which determine the outcome of actions. Paladyn . 2012 Dec 1;3(4):188-199. https://doi.org/10.2478/s13230-013-0104-x
Fichtl, Severin ; Alexander, John ; Kraft, Dirk ; Jorgensen, Jimmy Alison ; Kruger, Norbert ; Guerin, Frank. / Learning object relationships which determine the outcome of actions. In: Paladyn . 2012 ; Vol. 3, No. 4. pp. 188-199.
@article{b8e67461e0fb42c39b632d911f9c7016,
title = "Learning object relationships which determine the outcome of actions",
abstract = "Infants extend their repertoire of behaviours from initially simple behaviours with single objects to complex behaviours dealing with spatial relationships among objects. We are interested in the mechanisms underlying this development in order to achieve similar development in artificial systems. One mechanism is sensorimotor differentiation, which allows one behaviour to become altered in order to achieve a different result; the old behaviour is not forgotten, so differentiation increases the number of available behaviours. Differentiation requires the learning of both sensory abstractions and motor programs for the new behaviour; here we focus only on the sensory aspect: learning to recognise situations in which the new behaviour succeeds. We experimented with learning these situations in a realistic physical simulation of a robotic manipulator interacting with various objects, where the sensor space includes the robot arm position data and a Kinect-based vision system. The mechanism for learning sensory abstractions for a new behaviour is a component in the larger enterprise of building systems which emulate the mechanisms of infant development.",
keywords = "Developmental Artificial Intelligence, vision, infant development, means-end behaviour, learning preconditions",
author = "Severin Fichtl and John Alexander and Dirk Kraft and Jorgensen, {Jimmy Alison} and Norbert Kruger and Frank Guerin",
year = "2012",
month = "12",
day = "1",
doi = "10.2478/s13230-013-0104-x",
language = "English",
volume = "3",
pages = "188--199",
journal = "Paladyn",
issn = "2080-9778",
publisher = "De Gruyter Open Ltd.",
number = "4",

}

TY - JOUR

T1 - Learning object relationships which determine the outcome of actions

AU - Fichtl, Severin

AU - Alexander, John

AU - Kraft, Dirk

AU - Jorgensen, Jimmy Alison

AU - Kruger, Norbert

AU - Guerin, Frank

PY - 2012/12/1

Y1 - 2012/12/1

N2 - Infants extend their repertoire of behaviours from initially simple behaviours with single objects to complex behaviours dealing with spatial relationships among objects. We are interested in the mechanisms underlying this development in order to achieve similar development in artificial systems. One mechanism is sensorimotor differentiation, which allows one behaviour to become altered in order to achieve a different result; the old behaviour is not forgotten, so differentiation increases the number of available behaviours. Differentiation requires the learning of both sensory abstractions and motor programs for the new behaviour; here we focus only on the sensory aspect: learning to recognise situations in which the new behaviour succeeds. We experimented with learning these situations in a realistic physical simulation of a robotic manipulator interacting with various objects, where the sensor space includes the robot arm position data and a Kinect-based vision system. The mechanism for learning sensory abstractions for a new behaviour is a component in the larger enterprise of building systems which emulate the mechanisms of infant development.

AB - Infants extend their repertoire of behaviours from initially simple behaviours with single objects to complex behaviours dealing with spatial relationships among objects. We are interested in the mechanisms underlying this development in order to achieve similar development in artificial systems. One mechanism is sensorimotor differentiation, which allows one behaviour to become altered in order to achieve a different result; the old behaviour is not forgotten, so differentiation increases the number of available behaviours. Differentiation requires the learning of both sensory abstractions and motor programs for the new behaviour; here we focus only on the sensory aspect: learning to recognise situations in which the new behaviour succeeds. We experimented with learning these situations in a realistic physical simulation of a robotic manipulator interacting with various objects, where the sensor space includes the robot arm position data and a Kinect-based vision system. The mechanism for learning sensory abstractions for a new behaviour is a component in the larger enterprise of building systems which emulate the mechanisms of infant development.

KW - Developmental Artificial Intelligence

KW - vision

KW - infant development

KW - means-end behaviour

KW - learning preconditions

U2 - 10.2478/s13230-013-0104-x

DO - 10.2478/s13230-013-0104-x

M3 - Article

VL - 3

SP - 188

EP - 199

JO - Paladyn

JF - Paladyn

SN - 2080-9778

IS - 4

ER -