Advanced

Unsupervised learning of action primitives

Sanmohan, ; Krüger, Volker LU and Kragic, Danica (2010) 2010 10th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2010 p.554-559
Abstract

Action representation is a key issue in imitation learning for humanoids. With the recent finding of mirror neurons there has been a growing interest in expressing actions as a combination meaningful subparts called primitives. Primitives could be thought of as an alphabet for the human actions. In this paper we observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their effect on the involved object. While human movements can look vastly different even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. In this paper we propose an... (More)

Action representation is a key issue in imitation learning for humanoids. With the recent finding of mirror neurons there has been a growing interest in expressing actions as a combination meaningful subparts called primitives. Primitives could be thought of as an alphabet for the human actions. In this paper we observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their effect on the involved object. While human movements can look vastly different even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. In this paper we propose an unsupervised learning approach for action primitives that makes use of the human movements as well as the object state changes. We group actions according to the changes they make to the object state space. Movements that produce the same state change in the object state space are classified to be instances of the same action primitive. This allows us to define action primitives as sets of movements where the movements of each primitive are connected through the object state change they induce.

(Less)
Please use this url to cite or link to this publication:
author
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
host publication
2010 10th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2010
pages
6 pages
publisher
IEEE--Institute of Electrical and Electronics Engineers Inc.
conference name
2010 10th IEEE-RAS International Conference on Humanoid Robots, Humanoids 2010
conference location
Nashville, TN, United States
conference dates
2010-12-06 - 2010-12-08
external identifiers
  • scopus:79851504957
ISBN
9781424486885
9781424486908
DOI
10.1109/ICHR.2010.5686309
language
English
LU publication?
no
id
060e3765-64b2-4ea6-8d23-b626c00e50f5
date added to LUP
2019-06-28 09:23:14
date last changed
2019-07-26 02:18:59
@inproceedings{060e3765-64b2-4ea6-8d23-b626c00e50f5,
  abstract     = {<p>Action representation is a key issue in imitation learning for humanoids. With the recent finding of mirror neurons there has been a growing interest in expressing actions as a combination meaningful subparts called primitives. Primitives could be thought of as an alphabet for the human actions. In this paper we observe that human actions and objects can be seen as being intertwined: we can interpret actions from the way the body parts are moving, but as well from how their effect on the involved object. While human movements can look vastly different even under minor changes in location, orientation and scale, the use of the object can provide a strong invariant for the detection of motion primitives. In this paper we propose an unsupervised learning approach for action primitives that makes use of the human movements as well as the object state changes. We group actions according to the changes they make to the object state space. Movements that produce the same state change in the object state space are classified to be instances of the same action primitive. This allows us to define action primitives as sets of movements where the movements of each primitive are connected through the object state change they induce.</p>},
  author       = {Sanmohan,  and Krüger, Volker and Kragic, Danica},
  isbn         = {9781424486885},
  language     = {eng},
  location     = {Nashville, TN, United States},
  month        = {12},
  pages        = {554--559},
  publisher    = {IEEE--Institute of Electrical and Electronics Engineers Inc.},
  title        = {Unsupervised learning of action primitives},
  url          = {http://dx.doi.org/10.1109/ICHR.2010.5686309},
  year         = {2010},
}