王啸峰 (Jeff Wang)

Biography

I am Jeff Wang (王啸峰), currently serve as the deputy director of the Beijing Key Laboratory of General World Models. I received my Ph.D. from the Institute of Automation, Chinese Academy of Sciences (CASIA) in 2025. During my Ph.D., I published 20 papers in venues including CVPR, ICCV, ECCV, NeurIPS, ICLR, and AAAI. My work DriveDreamer was selected as one of the Most Influential ECCV'24 Papers, and I was invited by NVIDIA for an oral presentation. DriveDreamer has been deployed at most leading OEMs across China.

I work closely with Zheng Zhu. I lead frontier research and production deployment of world models. Our representative projects include DriveDreamer, GigaBrain, and GigaWorld. We are actively hiring full-time members and interns. Our members include recipients of prestigious programs like Huawei Genius Young Scholars (华为天才少年) and Alibaba Star (阿里星).

中文版本(点击展开)

我是王啸峰(Jeff)。2025 年博士毕业于中国科学院自动化研究所(CASIA),目前担任通用世界模型北京市重点实验室副主任。我关注物理世界的世界模型和VLA,面向具身智能与自动驾驶等应用方向。

博士期间,我在 CVPR、ICCV、ECCV、NeurIPS、ICLR、AAAI 等顶级会议与期刊发表论文 20 篇。代表性工作 DriveDreamer 入选 Most Influential ECCV'24 Papers,并受 NVIDIA 邀请进行口头报告;目前DriveDreamer已在国内多家头部主机厂落地应用。

我和朱政紧密合作。我的代表性项目包括 DriveDreamerGigaBrainGigaWorld。我们的团队成员来自于华为天才少年、阿里星等入选者。

Research Projects We are hiring outstanding full‑time members and interns to build next‑generation world models and embodied ai.

World model for embodied AI.

Paper Lists
Click to expand
News

VLA powered by world model.

Paper Lists
Click to expand
News
  • GigaBrain-0.5M* is released, achieving dozens of hours of zero-mistake clothes folding, coffee making, and box folding.
  • GigaBrain‑0.1 achieved 1st place on the RoboChallenge leaderboard (2026/02/09).
  • We will host GigaBrain Challenge @ CVPR 2026 workshop, see you in Denver.
  • GigaBrain‑0 code is released. [GitHub]

World model for autonomous driving.

Paper Lists
Click to expand
News
  • UniDriveDreamer & ReconDreamer-RL are accepted to ECCV 2026.
  • ReconDreamer++ is accepted to ICCV 2025.
  • DriveDreamer4D & ReconDreamer are accepted to CVPR 2025.
  • DriveDreamer‑2 is accepted to AAAI 2025.
  • DriveDreamer is selected as one of the Most Influential ECCV'24 Papers.
News
  • 2026-06: 7 papers are accepted to ECCV 2026.
  • 2026-04: 1 paper is accepted to ICML 2026.
  • 2026-02: 1 paper is accepted to CVPR 2026, 3 papers are recommended as findings.
  • 2026-01: 2 papers are accepted to ICLR 2026.
  • 2025-12: 2 papers are accepted to TPMAI.
  • 2025-09: 2 papers are accepted to NeurIPS 2026.
  • 2025-06: 3 papers are accepted to ICCV 2025.
  • 2025-03: 5 papers are accepted to CVPR 2025.
  • 2024-12: DriveDreamer-2 is accepted to AAAI 2025.
  • 2024-07: DriveDreamer is selected as one of the Most Influential ECCV'24 Papers.
  • 2024-07: DriveDreamer is accepted to ECCV 2024.
Show more
  • 2024-05: 1 paper on occupancy prediction is accepted to IJCAI 2024.
  • 2023-11: Excited to see Greg Brockman sharing thoughts on our ICLR'24 technique report!
  • 2023-07: 1 paper on 3D occupancy prediction is accepted to ICCV 2023.
  • 2023-02: 1 paper on 3D streaming perception is accepted to CVPR 2023.
  • 2023-01: 1 paper on 3D pretraining is accepted to ICLR 2023.
  • 2022-11: 1 paper on self-supervised depth estimation is accepted to AAAI 2023.
  • 2022-07: 1 paper on multi-view depth estimation is accepted to ECCV 2022.