A neural framework for robot motor learning based on memory consolidation

Hani Ben Amor, Shuhei Ikemoto, Takashi Minato, Bernhard Jung, Hiroshi Ishiguro

Research output: Chapter in Book/Report/Conference proceedingConference contribution

2 Citations (Scopus)

Abstract

Neural networks are a popular technique for learning the adaptive control of non-linear plants. When applied to the complex control of android robots, however, they suffer from serious limitations such as the moving target problem, i.e. the interference between old and newly learned knowledge. However, in order to achieve lifelong learning, it is important that robots are able to acquire new motor skills without for-getting previously learned ones. To overcome these problems, we propose a new framework for motor learning, which is based on consolidation. The framework contains a new rehearsal algorithm for retaining previously acquired knowledge and a growing neural network. In experiments, the framework was successfully applied to an artifical benchmark problem and a real-world android robot.

Original languageEnglish (US)
Title of host publicationLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Pages641-648
Number of pages8
Volume4432 LNCS
EditionPART 2
StatePublished - 2007
Externally publishedYes
Event8th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2007 - Warsaw, Poland
Duration: Apr 11 2007Apr 14 2007

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 2
Volume4432 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other8th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2007
CountryPoland
CityWarsaw
Period4/11/074/14/07

Fingerprint

Consolidation
Robot
Learning
Robots
Data storage equipment
Neural Networks
Life Long Learning
Neural networks
Growing Networks
Benchmarking
Motor Skills
Moving Target
Adaptive Control
Interference
Benchmark
Experiment
Framework
Memory Consolidation
Experiments
Knowledge

ASJC Scopus subject areas

  • Computer Science(all)
  • Biochemistry, Genetics and Molecular Biology(all)
  • Theoretical Computer Science

Cite this

Ben Amor, H., Ikemoto, S., Minato, T., Jung, B., & Ishiguro, H. (2007). A neural framework for robot motor learning based on memory consolidation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (PART 2 ed., Vol. 4432 LNCS, pp. 641-648). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4432 LNCS, No. PART 2).

A neural framework for robot motor learning based on memory consolidation. / Ben Amor, Hani; Ikemoto, Shuhei; Minato, Takashi; Jung, Bernhard; Ishiguro, Hiroshi.

Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4432 LNCS PART 2. ed. 2007. p. 641-648 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 4432 LNCS, No. PART 2).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Ben Amor, H, Ikemoto, S, Minato, T, Jung, B & Ishiguro, H 2007, A neural framework for robot motor learning based on memory consolidation. in Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 edn, vol. 4432 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), no. PART 2, vol. 4432 LNCS, pp. 641-648, 8th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2007, Warsaw, Poland, 4/11/07.
Ben Amor H, Ikemoto S, Minato T, Jung B, Ishiguro H. A neural framework for robot motor learning based on memory consolidation. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). PART 2 ed. Vol. 4432 LNCS. 2007. p. 641-648. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
Ben Amor, Hani ; Ikemoto, Shuhei ; Minato, Takashi ; Jung, Bernhard ; Ishiguro, Hiroshi. / A neural framework for robot motor learning based on memory consolidation. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics). Vol. 4432 LNCS PART 2. ed. 2007. pp. 641-648 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); PART 2).
@inproceedings{1f94e11a92934250b54fe05b12f56ff5,
title = "A neural framework for robot motor learning based on memory consolidation",
abstract = "Neural networks are a popular technique for learning the adaptive control of non-linear plants. When applied to the complex control of android robots, however, they suffer from serious limitations such as the moving target problem, i.e. the interference between old and newly learned knowledge. However, in order to achieve lifelong learning, it is important that robots are able to acquire new motor skills without for-getting previously learned ones. To overcome these problems, we propose a new framework for motor learning, which is based on consolidation. The framework contains a new rehearsal algorithm for retaining previously acquired knowledge and a growing neural network. In experiments, the framework was successfully applied to an artifical benchmark problem and a real-world android robot.",
author = "{Ben Amor}, Hani and Shuhei Ikemoto and Takashi Minato and Bernhard Jung and Hiroshi Ishiguro",
year = "2007",
language = "English (US)",
isbn = "9783540715900",
volume = "4432 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
number = "PART 2",
pages = "641--648",
booktitle = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
edition = "PART 2",

}

TY - GEN

T1 - A neural framework for robot motor learning based on memory consolidation

AU - Ben Amor, Hani

AU - Ikemoto, Shuhei

AU - Minato, Takashi

AU - Jung, Bernhard

AU - Ishiguro, Hiroshi

PY - 2007

Y1 - 2007

N2 - Neural networks are a popular technique for learning the adaptive control of non-linear plants. When applied to the complex control of android robots, however, they suffer from serious limitations such as the moving target problem, i.e. the interference between old and newly learned knowledge. However, in order to achieve lifelong learning, it is important that robots are able to acquire new motor skills without for-getting previously learned ones. To overcome these problems, we propose a new framework for motor learning, which is based on consolidation. The framework contains a new rehearsal algorithm for retaining previously acquired knowledge and a growing neural network. In experiments, the framework was successfully applied to an artifical benchmark problem and a real-world android robot.

AB - Neural networks are a popular technique for learning the adaptive control of non-linear plants. When applied to the complex control of android robots, however, they suffer from serious limitations such as the moving target problem, i.e. the interference between old and newly learned knowledge. However, in order to achieve lifelong learning, it is important that robots are able to acquire new motor skills without for-getting previously learned ones. To overcome these problems, we propose a new framework for motor learning, which is based on consolidation. The framework contains a new rehearsal algorithm for retaining previously acquired knowledge and a growing neural network. In experiments, the framework was successfully applied to an artifical benchmark problem and a real-world android robot.

UR - http://www.scopus.com/inward/record.url?scp=38049092662&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=38049092662&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9783540715900

VL - 4432 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 641

EP - 648

BT - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

ER -