Show simple item record

dc.contributor.author
Carius, Jan
dc.contributor.author
Farshidian, Farbod
dc.contributor.author
Hutter, Marco
dc.date.accessioned
2020-05-28T07:30:47Z
dc.date.available
2020-02-17T17:31:58Z
dc.date.available
2020-02-18T06:46:54Z
dc.date.available
2020-05-28T07:30:47Z
dc.date.issued
2020-04
dc.identifier.issn
2377-3766
dc.identifier.other
10.1109/LRA.2020.2974653
en_US
dc.identifier.uri
http://hdl.handle.net/20.500.11850/400020
dc.identifier.doi
10.3929/ethz-b-000400020
dc.description.abstract
We present an Imitation Learning approach for the control of dynamical systems with a known model. Our policy search method is guided by solutions from MPC. Typical policy search methods of this kind minimize a distance metric between the guiding demonstrations and the learned policy. Our loss function, however, corresponds to the minimization of the control Hamiltonian, which derives from the principle of optimality. Therefore, our algorithm directly attempts to solve the optimality conditions with a parameterized class of control laws. Additionally, the proposed loss function explicitly encodes the constraints of the optimal control problem and we provide numerical evidence that its minimization achieves improved constraint satisfaction. We train a mixture-of-expert neural network architecture for controlling a quadrupedal robot and show that this policy structure is well suited for such multimodal systems. The learned policy can successfully stabilize different gaits on the real walking robot from less than 10 min of demonstration data.
en_US
dc.format
application/pdf
en_US
dc.language.iso
en
en_US
dc.publisher
IEEE
en_US
dc.rights.uri
http://rightsstatements.org/page/InC-NC/1.0/
dc.subject
Learning from Demonstration
en_US
dc.subject
Legged Robots
en_US
dc.subject
Optimization and Optimal Control
en_US
dc.title
MPC-Net: A First Principles Guided Policy Search
en_US
dc.type
Journal Article
dc.rights.license
In Copyright - Non-Commercial Use Permitted
ethz.journal.title
IEEE Robotics and Automation Letters
ethz.journal.volume
5
en_US
ethz.journal.issue
2
en_US
ethz.pages.start
2897
en_US
ethz.pages.end
2904
en_US
ethz.size
8 p. accepted version
en_US
ethz.version.deposit
acceptedVersion
en_US
ethz.grant
Data-driven control approaches for advanced legged locomotion
en_US
ethz.grant
Perceptive Dynamic Locomotion on Rough Terrain
en_US
ethz.grant
subTerranean Haptic INvestiGator
en_US
ethz.identifier.wos
ethz.identifier.scopus
ethz.publication.place
Piscataway, NJ
en_US
ethz.publication.status
published
en_US
ethz.leitzahl
ETH Zürich::00002 - ETH Zürich::00012 - Lehre und Forschung::00007 - Departemente::02130 - Dep. Maschinenbau und Verfahrenstechnik / Dep. of Mechanical and Process Eng.::02620 - Inst. f. Robotik u. Intelligente Systeme / Inst. Robotics and Intelligent Systems::09570 - Hutter, Marco / Hutter, Marco
en_US
ethz.leitzahl.certified
ETH Zürich::00002 - ETH Zürich::00012 - Lehre und Forschung::00007 - Departemente::02130 - Dep. Maschinenbau und Verfahrenstechnik / Dep. of Mechanical and Process Eng.::02620 - Inst. f. Robotik u. Intelligente Systeme / Inst. Robotics and Intelligent Systems::09570 - Hutter, Marco / Hutter, Marco
en_US
ethz.tag
RSL
en_US
ethz.grant.agreementno
166232
ethz.grant.agreementno
188596
ethz.grant.agreementno
780883
ethz.grant.fundername
SNF
ethz.grant.fundername
SNF
ethz.grant.fundername
EC
ethz.grant.funderDoi
10.13039/501100001711
ethz.grant.funderDoi
10.13039/501100001711
ethz.grant.funderDoi
10.13039/501100000780
ethz.grant.program
H2020
ethz.grant.program
Projektförderung in Mathematik, Natur- und Ingenieurwissenschaften (Abteilung II)
ethz.grant.program
Projektförderung in Mathematik, Natur- und Ingenieurwissenschaften (Abteilung II)
ethz.date.deposited
2020-02-17T17:32:17Z
ethz.source
FORM
ethz.eth
yes
en_US
ethz.availability
Open access
en_US
ethz.rosetta.installDate
2020-05-28T07:31:01Z
ethz.rosetta.lastUpdated
2021-02-15T11:52:22Z
ethz.rosetta.versionExported
true
ethz.COinS
ctx_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.atitle=MPC-Net:%20A%20First%20Principles%20Guided%20Policy%20Search&rft.jtitle=IEEE%20Robotics%20and%20Automation%20Letters&rft.date=2020-04&rft.volume=5&rft.issue=2&rft.spage=2897&rft.epage=2904&rft.issn=2377-3766&rft.au=Carius,%20Jan&Farshidian,%20Farbod&Hutter,%20Marco&rft.genre=article&rft_id=info:doi/10.1109/LRA.2020.2974653&
 Search print copy at ETH Library

Files in this item

Thumbnail

Publication type

Show simple item record