SPADE: Sparse Pillar-based 3D Object Detection Accelerator for Autonomous Driving

3D object detection using point cloud (PC) data is essential for perception pipelines of autonomous driving, where efficient encoding is key to meeting stringent resource and latency requirements. PointPillars, a widely adopted bird's-eye view (BEV) encoding, aggregates 3D point cloud data into...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Proceedings - International Symposium on High-Performance Computer Architecture S. 454 - 467
Hauptverfasser: Lee, Minjae, Park, Seongmin, Kim, Hyungmin, Yoon, Minyong, Lee, Janghwan, Choi, Jun Won, Kim, Nam Sung, Kang, Mingu, Choi, Jungwook
Format: Tagungsbericht
Sprache:Englisch
Veröffentlicht: IEEE 02.03.2024
Schlagworte:
ISSN:2378-203X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Abstract 3D object detection using point cloud (PC) data is essential for perception pipelines of autonomous driving, where efficient encoding is key to meeting stringent resource and latency requirements. PointPillars, a widely adopted bird's-eye view (BEV) encoding, aggregates 3D point cloud data into 2D pillars for fast and accurate 3D object detection. However, the stateof-the-art methods employing PointPillars overlook the inherent sparsity of pillar encoding where only a valid pillar is encoded with a vector of channel elements, missing opportunities for significant computational reduction. Meanwhile, current sparse convolution accelerators are designed to handle only elementwise activation sparsity and do not effectively address the vector sparsity imposed by pillar encoding. In this paper, we propose SPADE, an algorithm-hardware codesign strategy to maximize vector sparsity in pillar-based 3D object detection and accelerate vector-sparse convolution commensurate with the improved sparsity. SPADE consists of three components: (1) a dynamic vector pruning algorithm balancing accuracy and computation savings from vector sparsity, (2) a sparse coordinate management hardware transforming 2D systolic array into a vector-sparse convolution accelerator, and (3) sparsityaware dataflow optimization tailoring sparse convolution schedules for hardware efficiency. Taped-out with a commercial technology, SPADE saves the amount of computation by 36.3-89.2% for representative 3D object detection networks and benchmarks, leading to 1.3-10.9 × speedup and 1.5-12.6 × energy savings compared to the ideal dense accelerator design. These sparsityproportional performance gains equate to 4.1-28.8 × speedup and 90.2-372.3 × energy savings compared to the counterpart server and edge platforms.
AbstractList 3D object detection using point cloud (PC) data is essential for perception pipelines of autonomous driving, where efficient encoding is key to meeting stringent resource and latency requirements. PointPillars, a widely adopted bird's-eye view (BEV) encoding, aggregates 3D point cloud data into 2D pillars for fast and accurate 3D object detection. However, the stateof-the-art methods employing PointPillars overlook the inherent sparsity of pillar encoding where only a valid pillar is encoded with a vector of channel elements, missing opportunities for significant computational reduction. Meanwhile, current sparse convolution accelerators are designed to handle only elementwise activation sparsity and do not effectively address the vector sparsity imposed by pillar encoding. In this paper, we propose SPADE, an algorithm-hardware codesign strategy to maximize vector sparsity in pillar-based 3D object detection and accelerate vector-sparse convolution commensurate with the improved sparsity. SPADE consists of three components: (1) a dynamic vector pruning algorithm balancing accuracy and computation savings from vector sparsity, (2) a sparse coordinate management hardware transforming 2D systolic array into a vector-sparse convolution accelerator, and (3) sparsityaware dataflow optimization tailoring sparse convolution schedules for hardware efficiency. Taped-out with a commercial technology, SPADE saves the amount of computation by 36.3-89.2% for representative 3D object detection networks and benchmarks, leading to 1.3-10.9 × speedup and 1.5-12.6 × energy savings compared to the ideal dense accelerator design. These sparsityproportional performance gains equate to 4.1-28.8 × speedup and 90.2-372.3 × energy savings compared to the counterpart server and edge platforms.
Author Yoon, Minyong
Kim, Hyungmin
Lee, Janghwan
Lee, Minjae
Park, Seongmin
Kim, Nam Sung
Choi, Jungwook
Kang, Mingu
Choi, Jun Won
Author_xml – sequence: 1
  givenname: Minjae
  surname: Lee
  fullname: Lee, Minjae
  email: lmj4666@hanyang.ac.kr
  organization: Hanyang University
– sequence: 2
  givenname: Seongmin
  surname: Park
  fullname: Park, Seongmin
  email: skstjdals@hanyang.ac.kr
  organization: Hanyang University
– sequence: 3
  givenname: Hyungmin
  surname: Kim
  fullname: Kim, Hyungmin
  email: kong4274@hanyang.ac.kr
  organization: Hanyang University
– sequence: 4
  givenname: Minyong
  surname: Yoon
  fullname: Yoon, Minyong
  email: ycivil93@hanyang.ac.kr
  organization: Hanyang University
– sequence: 5
  givenname: Janghwan
  surname: Lee
  fullname: Lee, Janghwan
  email: hwanii0288@hanyang.ac.kr
  organization: Hanyang University
– sequence: 6
  givenname: Jun Won
  surname: Choi
  fullname: Choi, Jun Won
  email: junwchoi@hanyang.ac.kr
  organization: Hanyang University
– sequence: 7
  givenname: Nam Sung
  surname: Kim
  fullname: Kim, Nam Sung
  email: nskim@illinois.edu
  organization: University of Illinois at Urbana-Champaign
– sequence: 8
  givenname: Mingu
  surname: Kang
  fullname: Kang, Mingu
  email: mingu@ucsd.edu
  organization: University of California,San Diego
– sequence: 9
  givenname: Jungwook
  surname: Choi
  fullname: Choi, Jungwook
  email: choij@hanyang.ac.kr
  organization: Hanyang University
BookMark eNotjs1Kw0AURkdRsK19gy7mBVLv_Gfchaa1QqGRKrgrk8yNTEmTMkkF396ALg5nc_j4puSu7VokZMFgyRjYp22xypTRSi45cLkEAMluyNwamwoFwgom-C2ZcGHShIP4fCDTvj-NGbeKTcjbocjy9TM9XFzskRahaVxMStejpyKn-_KE1UBzHEaFrqVZVWGD0Q1dpPVIdh26tjt3157mMXyH9uuR3Neu6XH-7xn52KzfV9tkt395XWW7JHCQQ1ICY95b5FhWVZ16LWrhtETlrbcO9Xidy8pzwV1pFHhtQNbeaSiZMMoxMSOLv92AiMdLDGcXf44MpNESlPgF8PRRcw
CODEN IEEPAD
ContentType Conference Proceeding
DBID 6IE
6IL
CBEJK
RIE
RIL
DOI 10.1109/HPCA57654.2024.00041
DatabaseName IEEE Electronic Library (IEL) Conference Proceedings
IEEE Proceedings Order Plan All Online (POP All Online) 1998-present by volume
IEEE Xplore All Conference Proceedings
IEEE Electronic Library (IEL)
IEEE Proceedings Order Plans (POP All) 1998-Present
DatabaseTitleList
Database_xml – sequence: 1
  dbid: RIE
  name: IEEE Electronic Library (IEL)
  url: https://ieeexplore.ieee.org/
  sourceTypes: Publisher
DeliveryMethod fulltext_linktorsrc
Discipline Computer Science
EISBN 9798350393132
EISSN 2378-203X
EndPage 467
ExternalDocumentID 10476405
Genre orig-research
GrantInformation_xml – fundername: National Research Foundation of Korea (NRF)
  funderid: 10.13039/501100003725
– fundername: Institute of Information & communications Technology Planning & Evaluation (IITP)
  grantid: IITP2024-RS-2023-00253914
  funderid: 10.13039/100019635
– fundername: Korea government (MSIT)
  grantid: 2022-0-00957
  funderid: 10.13039/501100014188
GroupedDBID 29O
6IE
6IF
6IH
6IK
6IL
6IM
6IN
AAJGR
AAWTH
ABLEC
ADZIZ
ALMA_UNASSIGNED_HOLDINGS
BEFXN
BFFAM
BGNUA
BKEBE
BPEOZ
CBEJK
CHZPO
IEGSK
IPLJI
M43
OCL
RIE
RIL
RNS
ID FETCH-LOGICAL-i204t-b011dd9e2ebccf8d63f3a64e5d9d9ae603924cd232ab750d6704fda60b1375a13
IEDL.DBID RIE
ISICitedReferencesCount 6
ISICitedReferencesURI http://www.webofscience.com/api/gateway?GWVersion=2&SrcApp=Summon&SrcAuth=ProQuest&DestLinkType=CitingArticles&DestApp=WOS_CPL&KeyUT=001207751400030&url=https%3A%2F%2Fcvtisr.summon.serialssolutions.com%2F%23%21%2Fsearch%3Fho%3Df%26include.ft.matches%3Dt%26l%3Dnull%26q%3D
IngestDate Wed Aug 27 02:09:38 EDT 2025
IsPeerReviewed false
IsScholarly true
Language English
LinkModel DirectLink
MergedId FETCHMERGED-LOGICAL-i204t-b011dd9e2ebccf8d63f3a64e5d9d9ae603924cd232ab750d6704fda60b1375a13
PageCount 14
ParticipantIDs ieee_primary_10476405
PublicationCentury 2000
PublicationDate 2024-March-2
PublicationDateYYYYMMDD 2024-03-02
PublicationDate_xml – month: 03
  year: 2024
  text: 2024-March-2
  day: 02
PublicationDecade 2020
PublicationTitle Proceedings - International Symposium on High-Performance Computer Architecture
PublicationTitleAbbrev HPCA
PublicationYear 2024
Publisher IEEE
Publisher_xml – name: IEEE
SSID ssj0002951
Score 2.3347068
Snippet 3D object detection using point cloud (PC) data is essential for perception pipelines of autonomous driving, where efficient encoding is key to meeting...
SourceID ieee
SourceType Publisher
StartPage 454
SubjectTerms Convolution
Encoding
Energy conservation
Heuristic algorithms
Object detection
Three-dimensional displays
Vectors
Title SPADE: Sparse Pillar-based 3D Object Detection Accelerator for Autonomous Driving
URI https://ieeexplore.ieee.org/document/10476405
WOSCitedRecordID wos001207751400030&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
link http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV09T8MwELWgYmAqH0V8ywOrqZ24dsIWtVQdUAkqoG6V7XOkLmmVJvx-bDctYmBgsyI5kc4638vdvXcIPUg-YCY2ijANCfGSb0RHsSFUaFkAJAICQ-7zRU6nyXye5i1ZPXBhrLWh-cw--mWo5cPKND5V1veyAoJ7xdJDKcWWrLW_diOHFVpuHKNpf5IPM4elQ94k4kGUk_2aoBICyLj7z0-foN4PFQ_n-yBzig5seYa6u1kMuHXNc_Q2y51Nn_Bs7f5U3RY_TKgiPkYBjkf4Vft8Cx7ZOrRelTgzxgWcUGPHDrfirKk9vWHVbPCoWvosQw99jJ_fhxPSjksgy4jymmjnqgCpjaw2pkhAxEWsBLcDSCFVVlAHhbgBB6GUdjgBhKS8ACWoZrEcKBZfoE65Ku0lwu4lCbCioKA1tylXzusLG0nBQGglzBXqeRMt1ltFjMXOOtd_PL9Bx_4UQu9WdIs6ddXYO3RkvurlproP5_gNt3Ke1g
linkProvider IEEE
linkToHtml http://cvtisr.summon.serialssolutions.com/2.0.0/link/0/eLvHCXMwlV07T8MwELYQIMFUHkW88cBqaieOk7BVLVURpQS1oG6V7XOkLmmVJvx-bPeBGBjYrEhOpLPO9-Xuvu8Quo95xHSoJWEKEuIk34gKQk2oUHEOkAjwDLnPQTwcJpNJmq3J6p4LY4zxzWfmwS19LR_munapspaTFRDcKZbuRZwHdEXX2l68gUULa3Yco2mrn3XaFk37zEnAvSwn-zVDxYeQXuOfHz9CzR8yHs62YeYY7ZjiBDU20xjw2jlP0fsos1Z9xKOF_Ve1W9w4oZK4KAU47OI35TIuuGsq33xV4LbWNuT4Kju2yBW368oRHOb1EnfLmcszNNFH72nc6ZP1wAQyCyiviLLOCpCawCit8wREmIdScBNBCqk0glowxDVYECWVRQogYspzkIIqFsaRZOEZ2i3mhTlH2L4kAZbnFJTiJuXS-n1uglgwEEoKfYGazkTTxUoTY7qxzuUfz-_QQX_8OpgOnocvV-jQnYjv5Aqu0W5V1uYG7euvarYsb_2ZfgMqQ6Id
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%3Abook&rft.genre=proceeding&rft.title=Proceedings+-+International+Symposium+on+High-Performance+Computer+Architecture&rft.atitle=SPADE%3A+Sparse+Pillar-based+3D+Object+Detection+Accelerator+for+Autonomous+Driving&rft.au=Lee%2C+Minjae&rft.au=Park%2C+Seongmin&rft.au=Kim%2C+Hyungmin&rft.au=Yoon%2C+Minyong&rft.date=2024-03-02&rft.pub=IEEE&rft.eissn=2378-203X&rft.spage=454&rft.epage=467&rft_id=info:doi/10.1109%2FHPCA57654.2024.00041&rft.externalDocID=10476405