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 language | English (US) |
---|---|
Title of host publication | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
Pages | 641-648 |
Number of pages | 8 |
Volume | 4432 LNCS |
Edition | PART 2 |
State | Published - 2007 |
Externally published | Yes |
Event | 8th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2007 - Warsaw, Poland Duration: Apr 11 2007 → Apr 14 2007 |
Publication series
Name | Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) |
---|---|
Number | PART 2 |
Volume | 4432 LNCS |
ISSN (Print) | 03029743 |
ISSN (Electronic) | 16113349 |
Other
Other | 8th International Conference on Adaptive and Natural Computing Algorithms, ICANNGA 2007 |
---|---|
Country | Poland |
City | Warsaw |
Period | 4/11/07 → 4/14/07 |
Fingerprint
ASJC Scopus subject areas
- Computer Science(all)
- Biochemistry, Genetics and Molecular Biology(all)
- Theoretical Computer Science
Cite this
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 proceeding › Conference contribution
}
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
AN - SCOPUS:38049092662
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 -