CusADi: A GPU Parallelization Framework for Symbolic Expressions and Optimal Control

The parallelism afforded by GPUs presents significant advantages in training controllers through reinforcement learning (RL). However, integrating model-based optimization into this process remains challenging due to the complexity of formulating and solving optimization problems across thousands of...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE robotics and automation letters Jg. 10; H. 2; S. 899 - 906
Hauptverfasser: Jeon, Se Hwan, Hong, Seungwoo, Lee, Ho Jae, Khazoom, Charles, Kim, Sangbae
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Piscataway IEEE 01.02.2025
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Schlagworte:
ISSN:2377-3766, 2377-3766
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract The parallelism afforded by GPUs presents significant advantages in training controllers through reinforcement learning (RL). However, integrating model-based optimization into this process remains challenging due to the complexity of formulating and solving optimization problems across thousands of instances. In this work, we present CusADi , an extension of the casadi symbolic framework to support the parallelization of arbitrary closed-form expressions on GPUs with CUDA . We also formulate a closed-form approximation for solving general optimal control problems, enabling large-scale parallelization and evaluation of MPC controllers. Our results show a ten-fold speedup relative to similar MPC implementation on the CPU, and we demonstrate the use of CusADi for various applications, including parallel simulation, parameter sweeps, and policy training.
AbstractList The parallelism afforded by GPUs presents significant advantages in training controllers through reinforcement learning (RL). However, integrating model-based optimization into this process remains challenging due to the complexity of formulating and solving optimization problems across thousands of instances. In this work, we present CusADi , an extension of the casadi symbolic framework to support the parallelization of arbitrary closed-form expressions on GPUs with CUDA . We also formulate a closed-form approximation for solving general optimal control problems, enabling large-scale parallelization and evaluation of MPC controllers. Our results show a ten-fold speedup relative to similar MPC implementation on the CPU, and we demonstrate the use of CusADi for various applications, including parallel simulation, parameter sweeps, and policy training.
Author Khazoom, Charles
Hong, Seungwoo
Kim, Sangbae
Jeon, Se Hwan
Lee, Ho Jae
Author_xml – sequence: 1
  givenname: Se Hwan
  orcidid: 0000-0002-2791-7850
  surname: Jeon
  fullname: Jeon, Se Hwan
  email: sehwan@mit.edu
  organization: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
– sequence: 2
  givenname: Seungwoo
  orcidid: 0000-0001-7838-3839
  surname: Hong
  fullname: Hong, Seungwoo
  email: swhong@mit.edu
  organization: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
– sequence: 3
  givenname: Ho Jae
  surname: Lee
  fullname: Lee, Ho Jae
  email: hjlee201@mit.edu
  organization: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
– sequence: 4
  givenname: Charles
  orcidid: 0000-0001-7224-1688
  surname: Khazoom
  fullname: Khazoom, Charles
  email: ckhaz@mit.edu
  organization: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
– sequence: 5
  givenname: Sangbae
  surname: Kim
  fullname: Kim, Sangbae
  email: sangbae@mit.edua
  organization: Department of Mechanical Engineering, Massachusetts Institute of Technology, Cambridge, MA, USA
BookMark eNpNkM1PwjAYhxuDiYjcPXho4nnYr63M2zIBTUggiuem694lw7HOdkTxr7cEDpze9_D83o_nFg1a2wJC95RMKCXp0_I9mzDCxITHlLFYXKEh41JGXCbJ4KK_QWPvt4QQGjPJ03iINvneZy_1M87wYv2J19rppoGm_tN9bVs8d3oHP9Z94co6_HHYFbapDZ79dg68D4THui3xquvrnW5wbtve2eYOXVe68TA-1xHazGeb_DVarhZvebaMDBNxHzHBTJpIrgHYVJCSpqwUwEhZpKmRodUJ4xUIalJiqORTEXPDiwKkIZoTPkKPp7Gds9978L3a2r1rw0bFqZDhRZYkgSInyjjrvYNKdS4c6w6KEnW0p4I9dbSnzvZC5OEUqQHgApdyKijh_8iDa7U
CODEN IRALC6
Cites_doi 10.1023/A:1008739131724
10.1142/S0219843615500395
10.1109/TRO.2015.2405592
10.1109/LRA.2023.3266985
10.1109/ICRA57147.2024.10610200
10.1017/CBO9780511804441
10.1109/lra.2024.3455907
10.1016/j.jpdc.2020.05.021
10.1007/s12532-020-00179-2
10.1109/Humanoids57100.2023.10375199
10.1126/scirobotics.adh5401
10.1007/978-3-030-44051-0_38
10.1007/s12532-018-0139-4
10.1137/S0363012902400713
10.1109/icra57147.2024.10611212
10.1007/s10514-013-9341-4
10.1109/LRA.2021.3057845
10.1109/IROS45743.2020.9341447
10.1109/HUMANOIDS.2017.8246929
10.1007/s10107-004-0559-y
10.1109/ICRA46639.2022.9811878
10.1007/978-1-4899-7560-7
10.1126/scirobotics.abk2822
10.1109/TRO.2023.3275384
10.1109/ICRA40945.2020.9196673
10.1109/LRA.2024.3407413
10.2140/pjm.1966.16.1
10.1007/0-387-30065-1_4
10.3182/20080706-5-KR-1001.01182
10.1109/ICRA46639.2022.9812369
10.1109/IROS.2009.5356084
ContentType Journal Article
Copyright Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2025
Copyright_xml – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2025
DBID 97E
RIA
RIE
AAYXX
CITATION
7SC
7SP
8FD
JQ2
L7M
L~C
L~D
DOI 10.1109/LRA.2024.3512254
DatabaseName IEEE All-Society Periodicals Package (ASPP) 2005–Present
IEEE All-Society Periodicals Package (ASPP) 1998–Present
IEEE Electronic Library (IEL)
CrossRef
Computer and Information Systems Abstracts
Electronics & Communications Abstracts
Technology Research Database
ProQuest Computer Science Collection
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts – Academic
Computer and Information Systems Abstracts Professional
DatabaseTitle CrossRef
Technology Research Database
Computer and Information Systems Abstracts – Academic
Electronics & Communications Abstracts
ProQuest Computer Science Collection
Computer and Information Systems Abstracts
Advanced Technologies Database with Aerospace
Computer and Information Systems Abstracts Professional
DatabaseTitleList
Technology Research Database
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Engineering
EISSN 2377-3766
EndPage 906
ExternalDocumentID 10_1109_LRA_2024_3512254
10778410
Genre orig-research
GroupedDBID 0R~
97E
AAJGR
AASAJ
AAWTH
ABQJQ
ABVLG
ACGFS
AGQYO
AGSQL
AHBIQ
AKJIK
AKQYR
ALMA_UNASSIGNED_HOLDINGS
ATWAV
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
EBS
EJD
IFIPE
IPLJI
JAVBF
KQ8
M43
M~E
O9-
OCL
RIA
RIE
AAYXX
CITATION
7SC
7SP
8FD
AARMG
ABAZT
JQ2
L7M
L~C
L~D
ID FETCH-LOGICAL-c245t-242c9673aee2840d192d4e20db99c7d4ea623fe41c90c1738453c3bbe7c0a303
IEDL.DBID RIE
ISICitedReferencesCount 2
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001381315200005&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
ISSN 2377-3766
IngestDate Mon Jun 30 12:59:12 EDT 2025
Thu Nov 27 01:04:43 EST 2025
Wed Dec 10 09:50:01 EST 2025
IsPeerReviewed true
IsScholarly true
Issue 2
Language English
License https://ieeexplore.ieee.org/Xplorehelp/downloads/license-information/IEEE.html
https://doi.org/10.15223/policy-029
https://doi.org/10.15223/policy-037
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-c245t-242c9673aee2840d192d4e20db99c7d4ea623fe41c90c1738453c3bbe7c0a303
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ORCID 0000-0001-7838-3839
0000-0001-7224-1688
0000-0002-2791-7850
PQID 3147527266
PQPubID 4437225
PageCount 8
ParticipantIDs crossref_primary_10_1109_LRA_2024_3512254
proquest_journals_3147527266
ieee_primary_10778410
PublicationCentury 2000
PublicationDate 2025-02-01
PublicationDateYYYYMMDD 2025-02-01
PublicationDate_xml – month: 02
  year: 2025
  text: 2025-02-01
  day: 01
PublicationDecade 2020
PublicationPlace Piscataway
PublicationPlace_xml – name: Piscataway
PublicationTitle IEEE robotics and automation letters
PublicationTitleAbbrev LRA
PublicationYear 2025
Publisher IEEE
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Publisher_xml – name: IEEE
– name: The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
References ref13
Chignoli (ref29) 2023
ref35
ref12
ref34
ref15
ref37
ref36
ref11
ref33
ref32
Bambade (ref39) 2023
ref2
ref1
ref17
ref16
Patterson (ref30) 2017
ref38
Bradbury (ref20) 2018
Plancher (ref14) 2019
Silver (ref48) 2018
Vingelmann (ref31) 2020
Paszke (ref21) 2019
Farshidian (ref8) 2023
Carpentier (ref28) 2019
ref46
ref23
Wchter (ref24) 2006; 106
ref45
ref26
ref25
ref47
Makoviychuk (ref22) 2021
ref42
ref41
ref44
Amos (ref19) 2017; 70
ref43
Gillis (ref10) 2020
ref27
Zhuang (ref3) 2023; 229
Hoeller (ref4) 2020
ref9
Lee (ref7) 2024
ref6
Kang (ref18) 2024
ref5
ref40
References_xml – ident: ref23
  doi: 10.1023/A:1008739131724
– volume-title: Computer Organization and Design RISC-V Edition: The Hardware Software Interface
  year: 2017
  ident: ref30
– volume: 70
  start-page: 136
  volume-title: Proc. 34th Int. Conf. Mach. Learn.
  year: 2017
  ident: ref19
  article-title: OptNet: Differentiable optimization as a layer in neural networks
– ident: ref41
  doi: 10.1142/S0219843615500395
– volume-title: Proc. 39th Benelux Meeting Syst. Control
  year: 2020
  ident: ref10
  article-title: Effortless modeling of optimal control problems with rockit
– volume-title: Proc. Toward Online Opt. Control Dyn. Robots Workshop Int. Conf. Robot. Automat.
  year: 2019
  ident: ref14
  article-title: Realtime model predictive control using parallel DDP on a GPU
– ident: ref44
  doi: 10.1109/TRO.2015.2405592
– ident: ref6
  doi: 10.1109/LRA.2023.3266985
– year: 2018
  ident: ref20
  article-title: JAX: Composable transformations of Python NumPy programs
– year: 2024
  ident: ref18
  article-title: Fast and certifiable trajectory optimization
– ident: ref2
  doi: 10.1109/ICRA57147.2024.10610200
– ident: ref33
  doi: 10.1017/CBO9780511804441
– ident: ref38
  doi: 10.1109/lra.2024.3455907
– ident: ref17
  doi: 10.1016/j.jpdc.2020.05.021
– ident: ref26
  doi: 10.1007/s12532-020-00179-2
– ident: ref34
  doi: 10.1109/Humanoids57100.2023.10375199
– ident: ref5
  doi: 10.1126/scirobotics.adh5401
– ident: ref15
  doi: 10.1007/978-3-030-44051-0_38
– ident: ref11
  doi: 10.1007/s12532-018-0139-4
– ident: ref35
  doi: 10.1137/S0363012902400713
– ident: ref16
  doi: 10.1109/icra57147.2024.10611212
– start-page: 990
  volume-title: Proc. Conf. Robot Learn.
  year: 2020
  ident: ref4
  article-title: Deep value model predictive control
– year: 2018
  ident: ref48
  article-title: Residual policy learning
– ident: ref42
  doi: 10.1007/s10514-013-9341-4
– ident: ref13
  doi: 10.1109/LRA.2021.3057845
– ident: ref40
  doi: 10.1109/IROS45743.2020.9341447
– ident: ref12
  doi: 10.1109/HUMANOIDS.2017.8246929
– volume: 106
  start-page: 25
  issue: 1
  year: 2006
  ident: ref24
  article-title: On the implementation of a primal-dual interior point filter line search algorithm for large-scale nonlinear programming
  publication-title: Math. Program.
  doi: 10.1007/s10107-004-0559-y
– ident: ref45
  doi: 10.1109/ICRA46639.2022.9811878
– year: 2021
  ident: ref22
  article-title: ISAAC gym: High performance GPU-based physics simulation for robot learning
  publication-title: NeurIPS Datasets Benchmarks
– ident: ref27
  doi: 10.1007/978-1-4899-7560-7
– year: 2023
  ident: ref39
  article-title: Proxqp: An efficient and versatile quadratic programming solver for real-time robotics applications and beyond
– start-page: 8026
  volume-title: Proc. 33rd Int. Conf. Neural Inf. Process. Syst.
  year: 2019
  ident: ref21
  article-title: Pytorch: An imperative style, high-performance deep learning library
– ident: ref1
  doi: 10.1126/scirobotics.abk2822
– year: 2020
  ident: ref31
  article-title: NVIDIA Cuda, release: 10.2.89
– ident: ref36
  doi: 10.1109/TRO.2023.3275384
– ident: ref9
  doi: 10.1109/ICRA40945.2020.9196673
– ident: ref37
  doi: 10.1109/LRA.2024.3407413
– ident: ref32
  doi: 10.2140/pjm.1966.16.1
– ident: ref25
  doi: 10.1007/0-387-30065-1_4
– start-page: 614
  volume-title: Proc. IEEE Int. Symp. Syst. Integrations
  year: 2019
  ident: ref28
  article-title: The pinocchio C libraryA fast and flexible implementation of rigid body dynamics algorithms and their analytical derivatives
– year: 2023
  ident: ref29
  article-title: Recursive rigid-body dynamics algorithms for systems with kinematic loops
– ident: ref46
  doi: 10.3182/20080706-5-KR-1001.01182
– year: 2024
  ident: ref7
  article-title: Integrating model-based footstep planning with model-free reinforcement learning for dynamic legged locomotion
– ident: ref47
  doi: 10.1109/ICRA46639.2022.9812369
– ident: ref43
  doi: 10.1109/IROS.2009.5356084
– volume: 229
  start-page: 73
  volume-title: Proc. Mach. Learn. Res.
  year: 2023
  ident: ref3
  article-title: Robot parkour learning
– year: 2023
  ident: ref8
  article-title: OCS2: An open source library for optimal control of switched systems
SSID ssj0001527395
Score 2.295082
Snippet The parallelism afforded by GPUs presents significant advantages in training controllers through reinforcement learning (RL). However, integrating model-based...
SourceID proquest
crossref
ieee
SourceType Aggregation Database
Index Database
Publisher
StartPage 899
SubjectTerms Closed form solutions
Codes
Controllers
Exact solutions
Graphics processing units
Optimal control
Optimization
Optimization and optimal control
Reinforcement learning
Robot programming
Software tools
software tools for robot programming
Trajectory
Title CusADi: A GPU Parallelization Framework for Symbolic Expressions and Optimal Control
URI https://ieeexplore.ieee.org/document/10778410
https://www.proquest.com/docview/3147527266
Volume 10
WOSCitedRecordID wos001381315200005&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
hasFullText 1
inHoldings 1
isFullTextHit
isPrint
journalDatabaseRights – providerCode: PRVIEE
  databaseName: IEEE Electronic Library (IEL)
  customDbUrl:
  eissn: 2377-3766
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0001527395
  issn: 2377-3766
  databaseCode: RIE
  dateStart: 20160101
  isFulltext: true
  titleUrlDefault: https://ieeexplore.ieee.org/
  providerName: IEEE
– providerCode: PRVHPJ
  databaseName: ROAD: Directory of Open Access Scholarly Resources
  customDbUrl:
  eissn: 2377-3766
  dateEnd: 99991231
  omitProxy: false
  ssIdentifier: ssj0001527395
  issn: 2377-3766
  databaseCode: M~E
  dateStart: 20160101
  isFulltext: true
  titleUrlDefault: https://road.issn.org
  providerName: ISSN International Centre
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELZoxQADzyIKpfLAwpDixk4cs0WlhQFKBUXqFtmOI1WiKeoDwcJv5-ykoggxsHmwFeu-2Hfnu_sOofMg8lmQEuNlVFlSbSk9pSXzNA2Nn2lHIeKaTfB-PxqNxKAsVne1MMYYl3xmWnboYvnpVC_tUxmccG7DZOChVzgPi2Kt7wcVSyUmglUokojLu8cYHECftShoNT9gP1SP66Xy6wJ2WqW3-8_97KGd0nzEcYH3Ptow-QHaXiMVPETDznIeX4-vcIxvBs94IGe2XcpLWW-Je6tsLAzmKn76mCjLDIy772VGbD7HMk_xA1wlE_hSp0hlr6Fhrzvs3Hpl7wRPg_AXNtKrRcipNAYUEEnBkEuZ8UmqhNAchhLsnsywthZEtzmNWEA1VcpwTSSotSNUzae5OUZYhZzDGpYRTZmQkfI1k5TJlEckVGlWRxcrqSavBUNG4jwLIhJAILEIJCUCdVSzUlybVwiwjhorHJLyDM0T2mYcMAUL4uSPZadoy7fteF0SdQNVF7OlOUOb-m0xns-aqHL_2W26n-QLLVa6bQ
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV09T8MwELVQQQIGPosoFPDAwpDixk4ds0WFUkQpFRSpW2Q7jlSJpqgfCP49ZzcRRYiBzUOsRPdi39l39x5C50HosyAhxkupsqTaUnpKS-Zp2jB-qh2FiBOb4N1uOBiIXt6s7nphjDGu-MzU7NDl8pOxnturMljh3KbJ4IS-aqWz8nat7ysVSyYmgiIZScRl5ymCI6DPahT8mh-wH87Hqan82oKdX2lt__OLdtBWHkDiaIH4Llox2R7aXKIV3Ef95nwaXQ-vcIRvey-4JydWMOU177jEraIeC0PAip8_R8pyA-Obj7wmNptimSX4ETaTEbypuShmL6N-66bfbHu5eoKnwfwzm-vVosGpNAZcEEkglEuY8UmihNAchhIin9SwuhZE1zkNWUA1VcpwTSQ4tgNUysaZOURYNTiHOSwlmjIhQ-VrJimTCQ9JQyVpBV0UVo3fFhwZsTtbEBEDArFFIM4RqKCyteLScwsDVlC1wCHOV9E0pnXGAVOIIY7-mHaG1tv9h07cueveH6MN34rzupLqKirNJnNzgtb0-2w4nZy6X-ULpZG8hQ
openUrl ctx_ver=Z39.88-2004&ctx_enc=info%3Aofi%2Fenc%3AUTF-8&rfr_id=info%3Asid%2Fsummon.serialssolutions.com&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Ajournal&rft.genre=article&rft.atitle=CusADi%3A+A+GPU+Parallelization+Framework+for+Symbolic+Expressions+and+Optimal+Control&rft.jtitle=IEEE+robotics+and+automation+letters&rft.au=Jeon%2C+Se+Hwan&rft.au=Hong%2C+Seungwoo&rft.au=Lee%2C+Ho+Jae&rft.au=Khazoom%2C+Charles&rft.date=2025-02-01&rft.issn=2377-3766&rft.eissn=2377-3766&rft.volume=10&rft.issue=2&rft.spage=899&rft.epage=906&rft_id=info:doi/10.1109%2FLRA.2024.3512254&rft.externalDBID=n%2Fa&rft.externalDocID=10_1109_LRA_2024_3512254
thumbnail_l http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/lc.gif&issn=2377-3766&client=summon
thumbnail_m http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/mc.gif&issn=2377-3766&client=summon
thumbnail_s http://covers-cdn.summon.serialssolutions.com/index.aspx?isbn=/sc.gif&issn=2377-3766&client=summon