DriveX
9th DriveX Workshop · IEEE/CVF CVPR 2027

Foundation Models for Autonomous Driving

A full-day CVPR forum on vision-language-action models, world models, 3D/4D spatial intelligence, cooperative autonomy, and safe deployment, with keynotes, panels, papers, live demos, and the eight-track DriveX Grand Challenge.

June 20 or June 21, 2027 Seattle, USA Full day · In person
  Call for Papers   Grand Challenge   Invited Speakers
VLA Models and World Models Reasoning, generative simulation, and closed-loop planning for driving.
Spatial and Cooperative Intelligence 3D/4D scene understanding and multi-agent V2X autonomy.
Trustworthy Deployment Safety, robustness, efficiency, and open benchmarks.
Workshop vision

Grounded, verifiable, and deployable embodied AI

DriveX@CVPR 2027 brings together the computer vision, robotics, and autonomous-driving communities to discuss foundation models for autonomous driving.

Autonomous driving is a large-scale embodied-AI problem: an agent must perceive, predict, reason, and act in uncertain 3D environments under strict safety and real-time constraints. Foundation models are shifting the field from isolated perception tasks toward unified systems that connect vision, language, geometry, time, and action. Driving Vision-Language-Action (VLA) models now couple reasoning with trajectory generation, generative world models have matured into action-conditioned simulators for closed-loop evaluation, and cooperative autonomy is moving toward shared world models.

The 9th edition of the DriveX series asks how learned representations can stay spatially and temporally grounded, how world models should support closed-loop planning and evaluation, how vehicles and infrastructure can share complementary evidence, and how reasoning, uncertainty, and efficiency make these systems trustworthy and deployable.

The full-day program features keynotes from academia and industry, two interactive panels with all invited speakers, archival and non-archival paper tracks, poster sessions with live demos presented by our sponsors, and the eight-track DriveX Grand Challenge. DriveX is part of the DriveX ecosystem.

Call for participation

Topics of interest

We welcome research spanning algorithms, systems, datasets, evaluation, and real-world deployment at the intersection of foundation models and autonomous driving.

Vision-Language-Action models and reasoning for driving
World models and generative closed-loop simulation
3D/4D spatial intelligence and reconstruction
Multi-agent and V2X cooperative autonomy
Long-tail reasoning, robustness, and uncertainty
Safety, verification, and closed-loop evaluation
Efficient on-vehicle deployment of foundation models
Sim-to-real transfer and synthetic data
Datasets, benchmarks, and agentic data curation
Language-conditioned human–AV interaction
Full-day program

Tentative schedule

Sixteen keynotes, two interactive panels, six paper orals, two poster sessions with live demos, and challenge winner talks. The final speaker order and room will be announced once CVPR confirms the workshop date.

Tentative full-day program for the DriveX workshop at CVPR 2027
TimeProgram
08:00–08:10Opening and DriveX ecosystem
08:10–09:30Keynotes I (4 talks)
09:30–10:00Paper orals I (3 talks)
10:00–10:20Coffee break and posters I
10:20–11:40Keynotes II (4 talks)
11:40–12:00Panel I: VLA and world models
12:00–13:00Lunch break
13:00–14:20Keynotes III (4 talks)
14:20–15:00Challenge winner talks
15:00–15:30Coffee, posters II, and sponsor live demos
15:30–16:50Keynotes IV (4 talks)
16:50–17:20Panel II: From benchmarks to safe deployment
17:20–17:50Paper orals II (3 talks)
17:50–18:00Awards and closing remarks

Panels are interactive: the audience votes on contested propositions before each panel, and moderators take questions from the room and an online queue. All talks are recorded via the CVPR Zoom room and, with the speakers' permission, published on YouTube. The schedule is tentative and subject to the final CVPR program.

Academic and industry perspectives

Invited speakers

We have invited leaders in 3D vision, world models, robot learning, foundation models, benchmarks, and production autonomy. The final speaker lineup will be announced on this page.

AGPortrait of Andreas Geiger

Andreas Geiger

University of Tübingen

Invited speaker
ADPortrait of Angela Dai

Angela Dai

Technical University of Munich

Invited speaker
DCPortrait of Daniel Cremers

Daniel Cremers

Technical University of Munich

Invited speaker
LLPortrait of Laura Leal-Taixé

Laura Leal-Taixé

NVIDIA

Invited speaker
MCPortrait of Manmohan Chandraker

Manmohan Chandraker

UC San Diego

Invited speaker
SLPortrait of Sharon Li

Sharon Li

University of Wisconsin–Madison

Invited speaker
BZPortrait of Bolei Zhou

Bolei Zhou

UCLA

Invited speaker
WBPortrait of Wolfram Burgard

Wolfram Burgard

University of Technology Nuremberg

Invited speaker
NRPortrait of Nicholas Roy

Nicholas Roy

MIT and Zoox

Invited speaker
MPPortrait of Marco Pavone

Marco Pavone

Stanford University and NVIDIA

Invited speaker
HCPortrait of Holger Caesar

Holger Caesar

TU Delft

Invited speaker
JMPortrait of Jiaqi Ma

Jiaqi Ma

UCLA

Invited speaker
JSPortrait of Jamie Shotton

Jamie Shotton

Wayve

Invited speaker
MTPortrait of Mingxing Tan

Mingxing Tan

Waymo

Invited speaker
AEPortrait of Ashok Elluswamy

Ashok Elluswamy

Tesla

Invited speaker
PMPortrait of Phil Michel

Phil Michel

Motional

Invited speaker
Call for papers

Archival and non-archival paper tracks

Submit full papers to the CVPR 2027 Workshop proceedings, or share extended abstracts and recently published work in the non-archival track. Reviewing on OpenReview is double-blind with at least three reviews per paper.

Archival

Full papers

  • CVPR 2027 paper style, up to 8 pages excluding references
  • Published in the CVPR 2027 Workshop proceedings (IEEE/CVF)
  • All authors need complete OpenReview profiles
Non-archival

Extended abstracts

  • Up to 4 pages, or recently published work
  • Posted on the workshop website, not in the proceedings

All accepted papers are presented as posters; six papers are selected for oral presentations. Authors of accepted papers will be invited to extend their work to journal length and submit it to a special issue organized by the DriveX organizers.

OpenReview link and detailed instructions coming soon

Three paper awards

Best Paper, Best Paper Runner-Up, and Best Industry Paper, nominated by the area chairs and presented in the closing session.

Program committee

Benchmark program

DriveX Grand Challenge: eight tracks

The tracks cover the stack from cooperative perception to reasoning, simulation, and closed-loop planning. Each track runs on a public Codabench leaderboard with a validation and a held-out test phase.

DriveX Grand Challenge tracks at CVPR 2027
TrackDatasetTaskMain metric
Track 1TUMTraf-V2XV2I cooperative 3D detection and trackingmAP3D, AMOTA
Track 2doScenesLanguage-instructed driving and navigationADE / FDE
Track 3MDriveClosed-loop multi-agent cooperative drivingDriving score
Track 4nuReasoningLong-tail driving reasoning with VLAsReasoning accuracy
Track 5HetroDHeterogeneous multi-agent traffic simulationRealism score
Track 6UrbanTwin–LUMPISim-to-real roadside LiDAR 3D detectionmAP3D
Track 7UrbanTwin–V2X-RealSim-to-real cooperative LiDAR 3D detectionmAP3D
Track 8Mixed SignalsV2X 3D detection with heterogeneous LiDARsmAP3D
Awards

Certificates and prizes

Winners receive challenge awards provided by our sponsors, including a certificate and a monetary prize, and present their solutions in the challenge session. All finalists are invited to the poster sessions.

  • All datasets are publicly available with dev kits and baselines
  • Prize eligibility requires a technical report and public code
  • Teams must disclose external data and pre-trained models
  • Participants are not required to cite papers beyond the dataset they use
Organizing team

Academia and industry, working together

36 organizers from 14 universities and 10 companies span computer vision, robotics, machine learning, intelligent transportation, and production autonomy.

WZPortrait of Walter Zimmer

Walter Zimmer

UCLA and TUM

General Chair · Challenge Chair
RSPortrait of Rui Song

Rui Song

UCLA and Cambridge

Program and Panel Chair
RGPortrait of Ross Greer

Ross Greer

UC Merced

Program and Panel Chair
ZHPortrait of Zhiyu Huang

Zhiyu Huang

UCLA

Program and Panel Chair
ZZPortrait of Zewei Zhou

Zewei Zhou

UCLA

Area Chair
HYPortrait of Haohan Yang

Haohan Yang

UCLA

Area Chair
TCPortrait of Tianhui Cai

Tianhui Cai

UCLA

Speaker Chair
SZPortrait of Seth Z. Zhao

Seth Z. Zhao

UCLA

Speaker Chair
YZPortrait of Yun Zhang

Yun Zhang

UCLA

Speaker Chair
CWPortrait of Chuheng Wei

Chuheng Wei

UC Riverside

Area Chair
HCPortrait of Hu Cao

Hu Cao

Southeast University

Web and Publicity
MRPortrait of Max Ronecker

Max Ronecker

NVIDIA

Industry and Sponsors
KLPortrait of Katie Luo

Katie Luo

Waymo

Industry and Sponsors
FPPortrait of Flavian Pegado

Flavian Pegado

Wayve

Industry and Sponsors
ZFPortrait of Zhe Fu

Zhe Fu

Stanford University

Diversity and Students
XCPortrait of Xiangyu Chen

Xiangyu Chen

NVIDIA

Industry and Sponsors
BLPortrait of Boyi Li

Boyi Li

NVIDIA

Diversity and Students
LWPortrait of Lena Wild

Lena Wild

KTH and TRATON

Diversity and Students
HYPortrait of Haibao Yu

Haibao Yu

Carnegie Mellon University

Area Chair
XZPortrait of Xingcheng Zhou

Xingcheng Zhou

Technical University of Munich

Web and Publicity
JZPortrait of Jiajie Zhang

Jiajie Zhang

Technical University of Munich

Web and Publicity
NPPortrait of Nils Purschke

Nils Purschke

UCLA and TUM

Web and Publicity
JLPortrait of Jack Langerman

Jack Langerman

Zoox

Industry and Sponsors
SPPortrait of Stephany Berrio Perez

Stephany Berrio Perez

University of Queensland

Diversity and Students
LUPortrait of Lars Ullrich

Lars Ullrich

NEURA Robotics

Area Chair
RJPortrait of Rajasekhar Josyula

Rajasekhar Josyula

Tesla

Industry and Sponsors
PLPortrait of Peizheng Li

Peizheng Li

University of Tübingen and Mercedes-Benz

Area Chair
SYPortrait of Senthil Yogamani

Senthil Yogamani

Qualcomm

Industry and Sponsors
TQPortrait of Tony (Xuewei) Qi

Tony (Xuewei) Qi

Motional

Advisory Board
DWPortrait of Daniel Watzenig

Daniel Watzenig

TU Graz

Advisory Board
HCPortrait of Holger Caesar

Holger Caesar

TU Delft

Challenge Chair
BZPortrait of Bolei Zhou

Bolei Zhou

UCLA

Advisory Board
MTPortrait of Mohan Trivedi

Mohan Trivedi

UC San Diego

Advisory Board
AKPortrait of Alois Knoll

Alois Knoll

Technical University of Munich

Advisory Board
JMPortrait of Jiaqi Ma

Jiaqi Ma

UCLA

Advisory Board
Workshop series

Previous DriveX editions

Explore programs, talks, papers, and community resources from previous DriveX workshops.

CVPR
2025 ICCV
2025 ITSC
2025 CVPR
2026 IV
2026 ECCV
2026 ITSC
2026 ICRA
2027