Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied

D. Jacob; D. Polani; C.L. Nehaniv

Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied

D. Jacob
, D. Polani
, C.L. Nehaniv

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

10 Citations (Scopus)

155 Downloads (Pure)

Abstract

Experiments to illustrate a novel methodology for
reinforcement learning in embodied physical agents are described.
A simulated legged robot is decomposed into structurebased
modules following the authors' EMBER principles of
local sensing, action and learning. The legs are individually
trained to 'walk' in isolation, and re-attached to the robot;
walking is then sufficiently stable that learning in situ can
continue. The experiments demonstrate the benefits of the
modular decomposition: state-space factorisation leads to faster
learning, in this case to the extent that an otherwise intractable
problem becomes learnable.

Original language	English
Title of host publication	Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation
Subtitle of host publication	CIRA 2005
Publisher	Institute of Electrical and Electronics Engineers (IEEE)
Pages	365-372
ISBN (Print)	0-7803-9355-4
Publication status	Published - 2005
Event	2005 IEEE Int Symposium of Computational Intelligence in Robotics & Automation - Espoo, Finland Duration: 27 Jun 2005 → 30 Jun 2005

Conference

Conference	2005 IEEE Int Symposium of Computational Intelligence in Robotics & Automation
Country/Territory	Finland
City	Espoo
Period	27/06/05 → 30/06/05

Access to Document

101964Final published version, 4.58 MB

Cite this

@inproceedings{8237327212154e3998e6e28739d8d543,

title = "Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied",

abstract = "Experiments to illustrate a novel methodology for reinforcement learning in embodied physical agents are described. A simulated legged robot is decomposed into structurebased modules following the authors' EMBER principles of local sensing, action and learning. The legs are individually trained to 'walk' in isolation, and re-attached to the robot; walking is then sufficiently stable that learning in situ can continue. The experiments demonstrate the benefits of the modular decomposition: state-space factorisation leads to faster learning, in this case to the extent that an otherwise intractable problem becomes learnable.",

author = "D. Jacob and D. Polani and C.L. Nehaniv",

year = "2005",

language = "English",

isbn = "0-7803-9355-4",

pages = "365--372",

booktitle = "Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation",

publisher = "Institute of Electrical and Electronics Engineers (IEEE)",

address = "United States",

note = "2005 IEEE Int Symposium of Computational Intelligence in Robotics \& Automation ; Conference date: 27-06-2005 Through 30-06-2005",

}

Jacob, D, Polani, D & Nehaniv, CL 2005, Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied. in Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation: CIRA 2005. Institute of Electrical and Electronics Engineers (IEEE), pp. 365-372, 2005 IEEE Int Symposium of Computational Intelligence in Robotics & Automation, Espoo, Finland, 27/06/05.

Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied. / Jacob, D.; Polani, D.; Nehaniv, C.L.
Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation: CIRA 2005. Institute of Electrical and Electronics Engineers (IEEE), 2005. p. 365-372.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

TY - GEN

T1 - Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied

AU - Jacob, D.

AU - Polani, D.

AU - Nehaniv, C.L.

PY - 2005

Y1 - 2005

N2 - Experiments to illustrate a novel methodology for reinforcement learning in embodied physical agents are described. A simulated legged robot is decomposed into structurebased modules following the authors' EMBER principles of local sensing, action and learning. The legs are individually trained to 'walk' in isolation, and re-attached to the robot; walking is then sufficiently stable that learning in situ can continue. The experiments demonstrate the benefits of the modular decomposition: state-space factorisation leads to faster learning, in this case to the extent that an otherwise intractable problem becomes learnable.

AB - Experiments to illustrate a novel methodology for reinforcement learning in embodied physical agents are described. A simulated legged robot is decomposed into structurebased modules following the authors' EMBER principles of local sensing, action and learning. The legs are individually trained to 'walk' in isolation, and re-attached to the robot; walking is then sufficiently stable that learning in situ can continue. The experiments demonstrate the benefits of the modular decomposition: state-space factorisation leads to faster learning, in this case to the extent that an otherwise intractable problem becomes learnable.

M3 - Conference contribution

SN - 0-7803-9355-4

SP - 365

EP - 372

BT - Procs 2005 IEEE Int Symposium on Computational Intelligence in Robotics and Automation

PB - Institute of Electrical and Electronics Engineers (IEEE)

T2 - 2005 IEEE Int Symposium of Computational Intelligence in Robotics & Automation

Y2 - 27 June 2005 through 30 June 2005

ER -

Legs that can walk: Embodiment-Based Modular Reinforcement Learning applied

Abstract

Conference

Access to Document

Fingerprint

Cite this