a qualitative-connectionist approach to Robotic Spatial Planning: the Peg-in-Hole case study

a qualitative-connectionist approach to Robotic Spatial Planning: the Peg-in-Hole case study Qualitative spatial reasoning foractual robots in real-world environments mustnecessarily involve perceptive knowledge, sincecomplete a-priori information of the outerworld can never be assumed, not even at aqualitative level. In this paper a contributionis made towards the integration of quantitativedata – namely, sensor signals – into a higherlevel qualitative plan. This integrationincludes the use of neural networks to learnhow to map complex perceptual signals into aqualitative description. Sensors are used toacquire actual knowledge of the environment andproperly identify, with the help of aconnectionist system, the real state of thesystem. The approach is presented in theframework of robotic tasks involving contactfor which the most informative perception comesfrom force/torque sensors. Empirical simulationresults are provided for the chamferlesstwo-dimensional peg-in-hole insertion modelwith friction. The advantages of learningapproaches over geometric model-basedtechniques are discussed: our approach issimple but robust against unpredictable changesof task parameters, and it exhibits agracefully degrading behavior and on-lineadaptation to new task conditions. Anenhancement to incorporate a measure ofconfidence of the network is also presented. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Spatial Cognition and Computation Springer Journals

a qualitative-connectionist approach to Robotic Spatial Planning: the Peg-in-Hole case study

Loading next page...
 
/lp/springer_journal/a-qualitative-connectionist-approach-to-robotic-spatial-planning-the-ulkIjEQyGD
Publisher
Springer Journals
Copyright
Copyright © 2000 by Kluwer Academic Publishers
Subject
Psychology; Cognitive Psychology
ISSN
1387-5868
eISSN
1573-9252
D.O.I.
10.1023/A:1011453027703
Publisher site
See Article on Publisher Site

Abstract

Qualitative spatial reasoning foractual robots in real-world environments mustnecessarily involve perceptive knowledge, sincecomplete a-priori information of the outerworld can never be assumed, not even at aqualitative level. In this paper a contributionis made towards the integration of quantitativedata – namely, sensor signals – into a higherlevel qualitative plan. This integrationincludes the use of neural networks to learnhow to map complex perceptual signals into aqualitative description. Sensors are used toacquire actual knowledge of the environment andproperly identify, with the help of aconnectionist system, the real state of thesystem. The approach is presented in theframework of robotic tasks involving contactfor which the most informative perception comesfrom force/torque sensors. Empirical simulationresults are provided for the chamferlesstwo-dimensional peg-in-hole insertion modelwith friction. The advantages of learningapproaches over geometric model-basedtechniques are discussed: our approach issimple but robust against unpredictable changesof task parameters, and it exhibits agracefully degrading behavior and on-lineadaptation to new task conditions. Anenhancement to incorporate a measure ofconfidence of the network is also presented.

Journal

Spatial Cognition and ComputationSpringer Journals

Published: Oct 16, 2004

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off