Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit

David Johan Christensen, Jørgen Christian Larsen, Kasper Støy

Research output: Contribution to conference without publisher/journalPaperResearchpeer-review

Abstract

This paper presents experiments with a morphologyindependent,
life-long strategy for online learning of locomotion
gaits, performed on a quadruped robot build using the LockKit
robotic building kit. The learning strategy applies a stochastic
optimization algorithm to optimize eight open parameters of a
central pattern generator based gait implementation. We observe
that the strategy converges in roughly ten minutes to gaits of
similar or higher velocity than a manually designed gait and
that the strategy readapts in the event of failed actuators. In
future work we plan to study co-learning of morphological and
control parameters directly on the physical robot.
Original languageEnglish
Publication date17. May 2012
Number of pages7
Publication statusPublished - 17. May 2012
EventIEEE Conference on Evolving and Adaptive Intelligent Systems - Carlos III University, Leganés Campus in Leganés, Madrid, Spain
Duration: 17. May 201218. May 2012
http://portal.uc3m.es/portal/page/portal/congresos_jornadas/home_cfp_eais

Conference

ConferenceIEEE Conference on Evolving and Adaptive Intelligent Systems
LocationCarlos III University, Leganés Campus in Leganés
CountrySpain
CityMadrid
Period17/05/201218/05/2012
Internet address

Cite this

Johan Christensen, D., Larsen, J. C., & Støy, K. (2012). Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit. Paper presented at IEEE Conference on Evolving and Adaptive Intelligent Systems, Madrid, Spain.
Johan Christensen, David ; Larsen, Jørgen Christian ; Støy, Kasper. / Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit. Paper presented at IEEE Conference on Evolving and Adaptive Intelligent Systems, Madrid, Spain.7 p.
@conference{4fab5907716a4fa9a6c94460bc572e52,
title = "Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit",
abstract = "This paper presents experiments with a morphologyindependent,life-long strategy for online learning of locomotiongaits, performed on a quadruped robot build using the LockKitrobotic building kit. The learning strategy applies a stochasticoptimization algorithm to optimize eight open parameters of acentral pattern generator based gait implementation. We observethat the strategy converges in roughly ten minutes to gaits ofsimilar or higher velocity than a manually designed gait andthat the strategy readapts in the event of failed actuators. Infuture work we plan to study co-learning of morphological andcontrol parameters directly on the physical robot.",
author = "{Johan Christensen}, David and Larsen, {J{\o}rgen Christian} and Kasper St{\o}y",
year = "2012",
month = "5",
day = "17",
language = "English",
note = "IEEE Conference on Evolving and Adaptive Intelligent Systems, EAIS ; Conference date: 17-05-2012 Through 18-05-2012",
url = "http://portal.uc3m.es/portal/page/portal/congresos_jornadas/home_cfp_eais",

}

Johan Christensen, D, Larsen, JC & Støy, K 2012, 'Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit', Paper presented at IEEE Conference on Evolving and Adaptive Intelligent Systems, Madrid, Spain, 17/05/2012 - 18/05/2012.

Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit. / Johan Christensen, David; Larsen, Jørgen Christian; Støy, Kasper.

2012. Paper presented at IEEE Conference on Evolving and Adaptive Intelligent Systems, Madrid, Spain.

Research output: Contribution to conference without publisher/journalPaperResearchpeer-review

TY - CONF

T1 - Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit

AU - Johan Christensen, David

AU - Larsen, Jørgen Christian

AU - Støy, Kasper

PY - 2012/5/17

Y1 - 2012/5/17

N2 - This paper presents experiments with a morphologyindependent,life-long strategy for online learning of locomotiongaits, performed on a quadruped robot build using the LockKitrobotic building kit. The learning strategy applies a stochasticoptimization algorithm to optimize eight open parameters of acentral pattern generator based gait implementation. We observethat the strategy converges in roughly ten minutes to gaits ofsimilar or higher velocity than a manually designed gait andthat the strategy readapts in the event of failed actuators. Infuture work we plan to study co-learning of morphological andcontrol parameters directly on the physical robot.

AB - This paper presents experiments with a morphologyindependent,life-long strategy for online learning of locomotiongaits, performed on a quadruped robot build using the LockKitrobotic building kit. The learning strategy applies a stochasticoptimization algorithm to optimize eight open parameters of acentral pattern generator based gait implementation. We observethat the strategy converges in roughly ten minutes to gaits ofsimilar or higher velocity than a manually designed gait andthat the strategy readapts in the event of failed actuators. Infuture work we plan to study co-learning of morphological andcontrol parameters directly on the physical robot.

M3 - Paper

ER -

Johan Christensen D, Larsen JC, Støy K. Adaptive Strategy for Online Gait Learning Evaluated on the Polymorphic Robotic LocoKit. 2012. Paper presented at IEEE Conference on Evolving and Adaptive Intelligent Systems, Madrid, Spain.