Apprentissage et optimisation de politiques pour un bras articulé actionné par des muscles
Identifieur interne : 001693 ( Main/Exploration ); précédent : 001692; suivant : 001694Apprentissage et optimisation de politiques pour un bras articulé actionné par des muscles
Auteurs : Didier Marin [France] ; Lionel Rigoux [France] ; Olivier Sigaud [France]Source :
- Revue d'intelligence artificielle [ 0992-499X ] ; 2013.
Descripteurs français
- Pascal (Inist)
- Wicri :
English descriptors
- KwdEn :
Abstract
Many research works combine learning from demonstration and policy improvement methods to learn the controller of a robot along a specific trajectory. Nevertheless, a capability to learn in the whole reachable space of this robot is missing in these works. In this paper we propose a method that consists in learning a reactive near-optimal feedback controller in two steps. First, an efficient parametric feedback controller is obtained from learning from Demonstration based on the trajectories computed by a costly near-optimal controller. Second, the feedback controller is optimized further with direct Policy Search methods. As a result, we obtain a controller that is executed 20 000 times faster than the original controller for a similar performance. Our work is evaluated in simulation.
Affiliations:
Links toward previous steps (curation, corpus...)
- to stream PascalFrancis, to step Corpus: 000062
- to stream PascalFrancis, to step Curation: 000945
- to stream PascalFrancis, to step Checkpoint: 000038
- to stream Main, to step Merge: 001709
- to stream Main, to step Curation: 001693
Le document en format XML
<record><TEI><teiHeader><fileDesc><titleStmt><title xml:lang="fr" level="a">Apprentissage et optimisation de politiques pour un bras articulé actionné par des muscles</title><author><name sortKey="Marin, Didier" sort="Marin, Didier" uniqKey="Marin D" first="Didier" last="Marin">Didier Marin</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author><author><name sortKey="Rigoux, Lionel" sort="Rigoux, Lionel" uniqKey="Rigoux L" first="Lionel" last="Rigoux">Lionel Rigoux</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author><author><name sortKey="Sigaud, Olivier" sort="Sigaud, Olivier" uniqKey="Sigaud O" first="Olivier" last="Sigaud">Olivier Sigaud</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author></titleStmt><publicationStmt><idno type="wicri:source">INIST</idno><idno type="inist">13-0216766</idno><date when="2013">2013</date><idno type="stanalyst">PASCAL 13-0216766 INIST</idno><idno type="RBID">Pascal:13-0216766</idno><idno type="wicri:Area/PascalFrancis/Corpus">000062</idno><idno type="wicri:Area/PascalFrancis/Curation">000945</idno><idno type="wicri:Area/PascalFrancis/Checkpoint">000038</idno><idno type="wicri:explorRef" wicri:stream="PascalFrancis" wicri:step="Checkpoint">000038</idno><idno type="wicri:doubleKey">0992-499X:2013:Marin D:apprentissage:et:optimisation</idno><idno type="wicri:Area/Main/Merge">001709</idno><idno type="wicri:Area/Main/Curation">001693</idno><idno type="wicri:Area/Main/Exploration">001693</idno></publicationStmt><sourceDesc><biblStruct><analytic><title xml:lang="fr" level="a">Apprentissage et optimisation de politiques pour un bras articulé actionné par des muscles</title><author><name sortKey="Marin, Didier" sort="Marin, Didier" uniqKey="Marin D" first="Didier" last="Marin">Didier Marin</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author><author><name sortKey="Rigoux, Lionel" sort="Rigoux, Lionel" uniqKey="Rigoux L" first="Lionel" last="Rigoux">Lionel Rigoux</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author><author><name sortKey="Sigaud, Olivier" sort="Sigaud, Olivier" uniqKey="Sigaud O" first="Olivier" last="Sigaud">Olivier Sigaud</name><affiliation wicri:level="3"><inist:fA14 i1="01"><s1>Institut des Systèmes lntelligents et de Robotique UPMC-Paris 6, CNRS UMR 7222 4 place Jussieu</s1><s2>75252 Paris</s2><s3>FRA</s3><sZ>1 aut.</sZ><sZ>2 aut.</sZ><sZ>3 aut.</sZ></inist:fA14><country>France</country><placeName><region type="region" nuts="2">Île-de-France</region><settlement type="city">Paris</settlement></placeName></affiliation></author></analytic><series><title level="j" type="main">Revue d'intelligence artificielle</title><title level="j" type="abbreviated">Rev. intell. artif.</title><idno type="ISSN">0992-499X</idno><imprint><date when="2013">2013</date></imprint></series></biblStruct></sourceDesc><seriesStmt><title level="j" type="main">Revue d'intelligence artificielle</title><title level="j" type="abbreviated">Rev. intell. artif.</title><idno type="ISSN">0992-499X</idno></seriesStmt></fileDesc><profileDesc><textClass><keywords scheme="KwdEn" xml:lang="en"><term>Arm</term><term>Capability index</term><term>Direct method</term><term>Entropy</term><term>Feedback regulation</term><term>Muscle</term><term>Optimal control</term><term>Optimal control (mathematics)</term><term>Optimization</term><term>Policy</term><term>Robotics</term><term>Search algorithm</term><term>Space application</term><term>Stochastic control</term></keywords><keywords scheme="Pascal" xml:lang="fr"><term>Robotique</term><term>Rétroaction</term><term>Politique</term><term>Bras</term><term>Muscle</term><term>Indice aptitude</term><term>Commande optimale</term><term>Optimisation</term><term>Application spatiale</term><term>Méthode directe</term><term>Algorithme recherche</term><term>Commande stochastique</term><term>Entropie</term><term>Contrôle optimal</term><term>.</term></keywords><keywords scheme="Wicri" type="topic" xml:lang="fr"><term>Robotique</term><term>Politique</term></keywords></textClass></profileDesc></teiHeader><front><div type="abstract" xml:lang="en">Many research works combine learning from demonstration and policy improvement methods to learn the controller of a robot along a specific trajectory. Nevertheless, a capability to learn in the whole reachable space of this robot is missing in these works. In this paper we propose a method that consists in learning a reactive near-optimal feedback controller in two steps. First, an efficient parametric feedback controller is obtained from learning from Demonstration based on the trajectories computed by a costly near-optimal controller. Second, the feedback controller is optimized further with direct Policy Search methods. As a result, we obtain a controller that is executed 20 000 times faster than the original controller for a similar performance. Our work is evaluated in simulation.</div></front></TEI><affiliations><list><country><li>France</li></country><region><li>Île-de-France</li></region><settlement><li>Paris</li></settlement></list><tree><country name="France"><region name="Île-de-France"><name sortKey="Marin, Didier" sort="Marin, Didier" uniqKey="Marin D" first="Didier" last="Marin">Didier Marin</name></region><name sortKey="Rigoux, Lionel" sort="Rigoux, Lionel" uniqKey="Rigoux L" first="Lionel" last="Rigoux">Lionel Rigoux</name><name sortKey="Sigaud, Olivier" sort="Sigaud, Olivier" uniqKey="Sigaud O" first="Olivier" last="Sigaud">Olivier Sigaud</name></country></tree></affiliations></record>Pour manipuler ce document sous Unix (Dilib)
EXPLOR_STEP=$WICRI_ROOT/Wicri/Lorraine/explor/InforLorV4/Data/Main/Exploration
HfdSelect -h $EXPLOR_STEP/biblio.hfd -nk 001693 | SxmlIndent | more
Ou
HfdSelect -h $EXPLOR_AREA/Data/Main/Exploration/biblio.hfd -nk 001693 | SxmlIndent | more
Pour mettre un lien sur cette page dans le réseau Wicri
{{Explor lien
|wiki= Wicri/Lorraine
|area= InforLorV4
|flux= Main
|étape= Exploration
|type= RBID
|clé= Pascal:13-0216766
|texte= Apprentissage et optimisation de politiques pour un bras articulé actionné par des muscles
}}
| This area was generated with Dilib version V0.6.33. |  |