DiNNO: Distributed Neural Network Optimization for Multi-Robot Collaborative Learning

We present DiNNO, a distributed algorithm that enables a group of robots to collaboratively optimize a deep neural network model while communicating over a mesh network. Each robot only has access to its own data and maintains its own version of the neural network, but eventually learns a model that...

Full description

Saved in:
Bibliographic Details
Published in:IEEE robotics and automation letters Vol. 7; no. 2; pp. 1896 - 1903
Main Authors: Yu, Javier, Vincent, Joseph A., Schwager, Mac
Format: Journal Article
Language:English
Published: Piscataway IEEE 01.04.2022
The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
Subjects:
ISSN:2377-3766, 2377-3766
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Abstract We present DiNNO, a distributed algorithm that enables a group of robots to collaboratively optimize a deep neural network model while communicating over a mesh network. Each robot only has access to its own data and maintains its own version of the neural network, but eventually learns a model that is as good as if it had been trained on all the data centrally. No robot sends raw data over the wireless network, preserving data privacy and ensuring efficient use of wireless bandwidth. At each iteration, each robot approximately optimizes an augmented Lagrangian function, then communicates the resulting weights to its neighbors, updates dual variables, and repeats. Eventually, all robots' local model weights reach a consensus. For convex objective functions, this consensus is a global optimum. Unlike many existing methods we test our algorithm on robotics-related, deep learning tasks with nontrivial model architectures. We compare DiNNO to two benchmark distributed deep learning algorithms in (i) an MNIST image classification task, (ii) a multi-robot implicit mapping task, and (iii) a multi-robot reinforcement learning task. In these experiments we show that DiNNO performs well when faced with nonconvex deep learning objectives, time-varying communication graphs, and streaming data. In all experiments our method outperforms baselines, and was able to achieve validation loss equivalent to centrally trained models. See msl.stanford.edu/projects/dist_nn_train for videos and code.
AbstractList We present DiNNO, a distributed algorithm that enables a group of robots to collaboratively optimize a deep neural network model while communicating over a mesh network. Each robot only has access to its own data and maintains its own version of the neural network, but eventually learns a model that is as good as if it had been trained on all the data centrally. No robot sends raw data over the wireless network, preserving data privacy and ensuring efficient use of wireless bandwidth. At each iteration, each robot approximately optimizes an augmented Lagrangian function, then communicates the resulting weights to its neighbors, updates dual variables, and repeats. Eventually, all robots’ local model weights reach a consensus. For convex objective functions, this consensus is a global optimum. Unlike many existing methods we test our algorithm on robotics-related, deep learning tasks with nontrivial model architectures. We compare DiNNO to two benchmark distributed deep learning algorithms in (i) an MNIST image classification task, (ii) a multi-robot implicit mapping task, and (iii) a multi-robot reinforcement learning task. In these experiments we show that DiNNO performs well when faced with nonconvex deep learning objectives, time-varying communication graphs, and streaming data. In all experiments our method outperforms baselines, and was able to achieve validation loss equivalent to centrally trained models. See msl.stanford.edu/projects/dist_nn_train for videos and code.
Author Yu, Javier
Vincent, Joseph A.
Schwager, Mac
Author_xml – sequence: 1
  givenname: Javier
  orcidid: 0000-0002-5552-8780
  surname: Yu
  fullname: Yu, Javier
  email: javieryu@stanford.edu
  organization: Department of Aeronautics and Astronautics, Stanford University, Stanford, CA, USA
– sequence: 2
  givenname: Joseph A.
  orcidid: 0000-0002-2270-7395
  surname: Vincent
  fullname: Vincent, Joseph A.
  email: josephav@stanford.edu
  organization: Department of Aeronautics and Astronautics, Stanford University, Stanford, CA, USA
– sequence: 3
  givenname: Mac
  orcidid: 0000-0002-7871-3663
  surname: Schwager
  fullname: Schwager, Mac
  email: schwager@stanford.edu
  organization: Department of Aeronautics and Astronautics, Stanford University, Stanford, CA, USA
BookMark eNp9kM1Lw0AQxRepYK29C14CnlP3I9nteiutXxBbKPW87CYT2Zpm62aj6F9vaouIB09vYN6bN_xOUa92NSB0TvCIECyvsuVkRDGlI0YSmmB6hPqUCREzwXnv13yChk2zxhiTlAom0z56mtn5fHEdzWwTvDVtgCKaQ-t11Ul4d_4lWmyD3dhPHayro9L56LGtgo2XzrgQTV1VaeN8t32DKAPta1s_n6HjUlcNDA86QKvbm9X0Ps4Wdw_TSRbnVJIQJ7gQuUzGAnReJiU34yLloFORUqNZoQFSrgupKSTESF5yWjImDJOGiZTlbIAu92e33r220AS1dq2vu0ZFOSVCjgmjnQvvXbl3TeOhVFtvN9p_KILVDp_q8KkdPnXA10X4n0huwzeA4LWt_gte7IMWAH56JO8eIZJ9AZpafqo
CODEN IRALC6
CitedBy_id crossref_primary_10_1109_LRA_2024_3379839
crossref_primary_10_32604_cmc_2023_033733
crossref_primary_10_1016_j_cviu_2023_103876
crossref_primary_10_3389_fnbot_2023_1329589
crossref_primary_10_1109_LRA_2024_3474551
crossref_primary_10_2514_1_I011337
crossref_primary_10_1007_s11424_024_3413_8
crossref_primary_10_1109_MRA_2024_3352852
crossref_primary_10_1109_MRA_2024_3358718
crossref_primary_10_1109_TCNS_2024_3487640
crossref_primary_10_3389_fnins_2023_1201370
crossref_primary_10_1109_TII_2024_3378839
crossref_primary_10_1109_ACCESS_2023_3264508
crossref_primary_10_1109_JSTSP_2024_3381373
crossref_primary_10_1109_LRA_2025_3575235
crossref_primary_10_3390_electronics12061402
crossref_primary_10_1007_s40747_023_01178_1
crossref_primary_10_1109_LRA_2025_3597513
crossref_primary_10_1109_TAI_2023_3247550
crossref_primary_10_32604_cmc_2023_041897
crossref_primary_10_1109_TIV_2023_3293198
crossref_primary_10_1007_s11431_024_2680_3
Cites_doi 10.1007/978-3-030-20205-7_3
10.1137/14096668X
10.1007/978-3-319-71682-4_5
10.1109/TAC.2008.2009515
10.1007/978-3-030-58452-8_24
10.1145/3394486.3403109
10.23919/ACC.2017.7962962
10.12681/eadd/3778
10.1109/DSW.2019.8755807
10.1109/ICCV48922.2021.00617
10.1109/MSP.2020.2970170
10.1007/978-3-030-05816-6_3
10.1561/9781601984616
10.1109/IROS.2017.8202141
10.1109/JPROC.2018.2817461
10.1109/ICRA.2018.8460473
10.1109/ICRA48506.2021.9560791
10.1109/TSP.2015.2436358
10.1109/MLSP.2016.7738894
10.1109/TSP.2014.2304432
10.1007/978-3-319-46128-1_50
10.1109/LRA.2020.3048652
10.1109/TAC.2016.2525928
10.1109/TSP.2010.2055862
10.1109/ICRA.2017.7989037
10.1109/TRO.2021.3098436
10.1109/TSP.2014.2367458
10.1007/s10107-020-01487-0
10.1007/978-94-6265-282-8
10.1109/TAC.2014.2364096
10.1016/j.arcontrol.2019.05.006
ContentType Journal Article
Copyright Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2022
Copyright_xml – notice: Copyright The Institute of Electrical and Electronics Engineers, Inc. (IEEE) 2022
DBID 97E
RIA
RIE
AAYXX
CITATION
7SC
7SP
8FD
JQ2
L7M
L~C
L~D
DOI 10.1109/LRA.2022.3142402
DatabaseName IEEE All-Society Periodicals Package (ASPP) 2005–Present
IEEE All-Society Periodicals Package (ASPP) 1998–Present
IEEE/IET Electronic Library (IEL) (UW System Shared)
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 1903
ExternalDocumentID 10_1109_LRA_2022_3142402
9681319
Genre orig-research
GrantInformation_xml – fundername: NSF Graduate Research Fellowship
– fundername: NASA ULI
  grantid: 80NSSC20M0163
– fundername: NSF NRI
  grantid: 1925030; 1830402
– fundername: Dwight D. Eisenhower Transportation Fellowship
GroupedDBID 0R~
97E
AAJGR
AARMG
AASAJ
AAWTH
ABAZT
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
JQ2
L7M
L~C
L~D
ID FETCH-LOGICAL-c291t-40d7c9487eacf4f6b8d56ea5752ba3daee56ad9a2e41b96f62f337b39b3753c3
IEDL.DBID RIE
ISICitedReferencesCount 21
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=000745524600020&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 Sun Nov 09 06:57:44 EST 2025
Tue Nov 18 22:25:57 EST 2025
Sat Nov 29 06:03:14 EST 2025
Wed Aug 27 03:02:26 EDT 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-c291t-40d7c9487eacf4f6b8d56ea5752ba3daee56ad9a2e41b96f62f337b39b3753c3
Notes ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ORCID 0000-0002-7871-3663
0000-0002-5552-8780
0000-0002-2270-7395
PQID 2621798132
PQPubID 4437225
PageCount 8
ParticipantIDs proquest_journals_2621798132
crossref_citationtrail_10_1109_LRA_2022_3142402
crossref_primary_10_1109_LRA_2022_3142402
ieee_primary_9681319
PublicationCentury 2000
PublicationDate 2022-04-01
PublicationDateYYYYMMDD 2022-04-01
PublicationDate_xml – month: 04
  year: 2022
  text: 2022-04-01
  day: 01
PublicationDecade 2020
PublicationPlace Piscataway
PublicationPlace_xml – name: Piscataway
PublicationTitle IEEE robotics and automation letters
PublicationTitleAbbrev LRA
PublicationYear 2022
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 ref35
ref34
ref15
ref37
Paszke (ref2) 2019
ref14
LeCun (ref36) 1998
ref31
ref30
ref11
ref33
ref10
Koloskova (ref24) 2020
ref32
Zhang (ref13) 2018
Corder (ref7) 2019
ref1
Terry (ref42) 2020
Schulman (ref45) 2017
ref17
ref39
ref16
ref38
ref19
Tancik (ref40) 2020
ref18
Luo (ref12) 2019
ref46
Halsted (ref20) 2021
ref26
ref25
ref22
ref21
ref28
ref27
Lian (ref23) 2017
ref29
ref8
Lowe (ref41) 2017
ref9
ref4
Kingma (ref3) 2014
ref6
ref5
Papoudakis (ref43) 2021
Terry (ref44) 2020
References_xml – ident: ref39
  doi: 10.1007/978-3-030-20205-7_3
– ident: ref17
  doi: 10.1137/14096668X
– ident: ref46
  doi: 10.1007/978-3-319-71682-4_5
– start-page: 8024
  volume-title: Proc. 33rd Int. Conf. Neural Inf. Process. Syst.
  year: 2019
  ident: ref2
  article-title: PyTorch: An imperative style, high-performance deep learning library
– year: 2021
  ident: ref20
  article-title: A survey of distributed optimization methods for multi-robot systems
– start-page: 5336
  volume-title: Proc. 31st Int. Conf. Neural Inf. Process. Syst.
  year: 2017
  ident: ref23
  article-title: Can decentralized algorithms outperform centralized algorithms? A case study for decentralized parallel stochastic gradient descent
– ident: ref16
  doi: 10.1109/TAC.2008.2009515
– year: 2019
  ident: ref7
  article-title: Decentralized multi-agent actor-critic with generative inference
– ident: ref37
  doi: 10.1007/978-3-030-58452-8_24
– ident: ref28
  doi: 10.1145/3394486.3403109
– ident: ref35
  doi: 10.23919/ACC.2017.7962962
– ident: ref15
  doi: 10.12681/eadd/3778
– ident: ref27
  doi: 10.1109/DSW.2019.8755807
– ident: ref38
  doi: 10.1109/ICCV48922.2021.00617
– year: 2020
  ident: ref42
  article-title: Revisiting parameter sharing in multi-agent deep reinforcement learning
– year: 2020
  ident: ref44
  article-title: PettingZoo: Gym for multi-agent reinforcement learning
– volume-title: Proc. Int. Conf. Neural Inf. Process. Syst.
  year: 2020
  ident: ref40
  article-title: Fourier features let networks learn high frequency functions in low dimensional domains
– ident: ref30
  doi: 10.1109/MSP.2020.2970170
– ident: ref9
  doi: 10.1007/978-3-030-05816-6_3
– ident: ref1
  doi: 10.1561/9781601984616
– ident: ref8
  doi: 10.1109/IROS.2017.8202141
– volume-title: Proc. 35th Conf. Neural Inf. Process. Syst. Datasets Benchmarks Track
  year: 2021
  ident: ref43
  article-title: Benchmarking multi-agent deep reinforcement learning algorithms in cooperative tasks
– ident: ref21
  doi: 10.1109/JPROC.2018.2817461
– ident: ref10
  doi: 10.1109/ICRA.2018.8460473
– year: 1998
  ident: ref36
  article-title: The mnist database of handwritten digits
– ident: ref14
  doi: 10.1109/ICRA48506.2021.9560791
– ident: ref29
  doi: 10.1109/TSP.2015.2436358
– ident: ref25
  doi: 10.1109/MLSP.2016.7738894
– ident: ref34
  doi: 10.1109/TSP.2014.2304432
– ident: ref33
  doi: 10.1007/978-3-319-46128-1_50
– ident: ref11
  doi: 10.1109/LRA.2020.3048652
– ident: ref19
  doi: 10.1109/TAC.2016.2525928
– ident: ref31
  doi: 10.1109/TSP.2010.2055862
– ident: ref6
  doi: 10.1109/ICRA.2017.7989037
– ident: ref5
  doi: 10.1109/TRO.2021.3098436
– volume-title: Proc. Int. Conf. Learn. Representations
  year: 2020
  ident: ref24
  article-title: Decentralized deep learning with arbitrary communication compression
– start-page: 1488
  volume-title: Proc. 18th Int. Conf. Auton. Agents MultiAgent Syst.
  year: 2019
  ident: ref12
  article-title: Distributed environmental modeling and adaptive sampling for multi-robot sensor coverage
– start-page: 6382
  volume-title: Proc. 31st Int. Conf. Neural Inf. Process. Syst.
  year: 2017
  ident: ref41
  article-title: Multi-agent actor-critic for mixed cooperative-competitive environments
– ident: ref32
  doi: 10.1109/TSP.2014.2367458
– ident: ref26
  doi: 10.1007/s10107-020-01487-0
– year: 2017
  ident: ref45
  article-title: Proximal policy optimization algorithms
– year: 2014
  ident: ref3
  article-title: Adam: A method for stochastic optimization
– start-page: 5872
  volume-title: Proc. 35th Int. Conf. Mach. Learn.
  year: 2018
  ident: ref13
  article-title: Fully decentralized multi-agent reinforcement learning with networked agents
– ident: ref4
  doi: 10.1007/978-94-6265-282-8
– ident: ref18
  doi: 10.1109/TAC.2014.2364096
– ident: ref22
  doi: 10.1016/j.arcontrol.2019.05.006
SSID ssj0001527395
Score 2.4221
Snippet We present DiNNO, a distributed algorithm that enables a group of robots to collaboratively optimize a deep neural network model while communicating over a...
SourceID proquest
crossref
ieee
SourceType Aggregation Database
Enrichment Source
Index Database
Publisher
StartPage 1896
SubjectTerms Algorithms
Artificial neural networks
Cognitive tasks
Communication
Data models
Deep learning
Deep learning methods
distributed robot systems
Finite element method
Image classification
Iterative methods
Lagrangian function
Machine learning
multi-robot systems
Multiple robots
Network management systems
Neural networks
Optimization
Robotics
Robots
Task analysis
Training
Wireless networks
Title DiNNO: Distributed Neural Network Optimization for Multi-Robot Collaborative Learning
URI https://ieeexplore.ieee.org/document/9681319
https://www.proquest.com/docview/2621798132
Volume 7
WOSCitedRecordID wos000745524600020&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/eLvHCXMwlV07T8MwED61FQMMvAqiUCoPLEiEtnk4MVvVhxggRVWRukWxfUGVoEV9jfx2zknagkBIbBlsK_Jn--7z-b4DuNKBJr9DepaUJswYCLRo3UhLxxJjMleeUiotNuGHYTAaiacC3GxyYRAxfXyGt-YzjeXrqVqaq7K64EHTMRqfRd_nWa7W9j7FKIkJbx2JbIj6w6BF_M-2iZa6JobwzfKkpVR-nL-pUekd_O93DmE_dx5ZK0P7CAo4OYa9L5KCZXjujMOwf8c6RhDX1LJCzYwAB3ULsxffrE-nxFuefsnIZ2VpEq41mMrpgrW362KFLFdffTmBYa87bN9beekES9miuSBWqH0liIzQuZq4CZeB9jjG5JvZMnZ0jOjxWIvYRrcpBU-4nTiOT3BJh_iLck6hNJlO8AyYh66vm7F0uUyMsRfkD2pf8gAJYRqrAvX1rEYqlxU31S1eo5ReNEREOEQGhyjHoQLXmx7vmaTGH23LZt437fIpr0B1DVyU77l5ZHPbqK8RvT7_vdcF7Jqxs3c3VSgtZku8hB21WoznsxoUHz-6tXRRfQKVd8th
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV3fS8MwED7GFNQHf01x_syDL4J1a9qmjW-yKYqzkzFhb6VJrjLQTeb07_fSdlNRBN_6kKQlX5q7L5f7DuDYRIb8DhU4StkwYyTRoXWjHJMqTMlcBVrrvNhEGMfRYCDvK3A6z4VBxPzyGZ7ZxzyWb8b6zR6VNaSIXM9qfC4Evs-bRbbW54mK1RKTwSwW2ZSNTu-CGCDnREx9G0X4ZnvyYio_duDcrFyt_e-D1mG1dB_ZRYH3BlRwtAkrX0QFa_DQHsZx95y1rSSurWaFhlkJDuoWF3e-WZf2iecyAZOR18ryNFynN1bjKWt9rox3ZKX-6uMW9K8u-61rpyye4Ggu3SnxQhNqSXSEdtbMz4SKTCAwJe-Mq9QzKWIgUiNTjr6rpMgEzzwvJMCURwxGe9tQHY1HuAMsQD80bqp8oTJr7iV5hCZUIkLCmMaqQ2M2q4kuhcVtfYunJCcYTZkQDonFISlxqMPJvMdLIarxR9uanfd5u3LK67A_Ay4p_7rXhAtu9deIYO_-3usIlq77d52kcxPf7sGyfU9xC2cfqtPJGx7Aon6fDl8nh_nS-gCBMM13
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=DiNNO%3A+Distributed+Neural+Network+Optimization+for+Multi-Robot+Collaborative+Learning&rft.jtitle=IEEE+robotics+and+automation+letters&rft.au=Yu%2C+Javier&rft.au=Vincent%2C+Joseph+A.&rft.au=Schwager%2C+Mac&rft.date=2022-04-01&rft.pub=IEEE&rft.eissn=2377-3766&rft.volume=7&rft.issue=2&rft.spage=1896&rft.epage=1903&rft_id=info:doi/10.1109%2FLRA.2022.3142402&rft.externalDocID=9681319
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