Shiyu Song's Homepage

The homepage has been moved to https://songshiyu01.github.io/.

 Brief Biography

    Senior Research Scientist (Tech lead), Baidu ADU, Sunnyvale, CA USA (Nov. 2014 - present), We're hiring!
    I'm one of the founding team members of the Baidu Autonomous Driving Car project. I joined Baidu in 2014. I'm a senior research scientist at Baidu Autonomous Driving Business Unit (ADU) now. Since Jan. 2016, I am the tech lead of the vehicle localization and high-definition (HD) map team at Baidu ADU.

    Doctor of Philosophy in University of California, San Diego.
    My research interests include Computer Vision, Robotics, LiDAR/Camera based Vehicle Localization, Visual Perception, Visual Odometry (VO), Simultaneous Localization and Mapping (SLAM), Structure from Motion (SFM), 3D Reconstruction.
Contact
        Email: songshiyu [AT] baidu.com
News
  • A paper is accepted in ICRA 2018 (01/16/2018).
  • We published a paper introducing our multi-sensor fusion based localization system (to be released in Apollo 2.0) in arXiv (11/15/2017).
  • I gave a keynote speech at the Laser Scanning Workshop in conjunction with the ISPRS Geospatial Week 2017 (09/19/2017).
  • Our team received the Baidu Best Team award in Wuzhen project (12/2016).
  • Xinwen Lianbo spent 2min30s introducing Baidu driverless car! Check here (12/10/2015).
  • Baidu driverless car completes test on Beijing roads (I feel very proud that the localization technology is developed by me) Check here (12/10/2015).
  • Received Award Employee of the Quarter at Baidu IDL US (Q3 2015).
  • Patent US-9,148,650 is grant (09/29/2015).
  • A paper is accepted by T-PAMI (07/31/2015).
  • Patent US-9,070,202 is grant (06/31/2015).
  • A paper is accepted in CVPR 2015 (Oral) (03/02/2015).
Awards
  • Baidu Best Team, 12/2016
  • Employee of the Quarter, Baidu IDL US, Q3 2015
  • Spot Recognition Award, "Real-Time Monocular SFM for Autonomous Driving", Department of Media Analytics, NEC Laboratories America, 2012
Projects and Publications
Guowei Wan, Xiaolong Yang, Renlan Cai, Hao Li, Hao Wang, Shiyu Song, "Robust and Precise Vehicle Localization based on Multi-sensor Fusion in Diverse City Scenes", Published in arXiv, Nov 15, 2017, [link][pdf][video]
Guowei Wan, Xiaolong Yang, Renlan Cai, Hao Li, Hao Wang, Shiyu Song, "Robust and Precise Vehicle Localization based on Multi-sensor Fusion in Diverse City Scenes", Accepted in ICRA 2018IEEE International Conference on Robotics and Automation, Brisbane, Australia, May 21-25, 2018




Shiyu Song, Manmohan Chandraker, "Joint SFM and Detection Cues for Monocular 3D Localization in Road Scenes", In CVPR 2015: IEEE Conference on Computer Vision and Pattern Recognition, Boston, Massachusetts, June 8-10, 2015, [pdf]  (Oral presentation Acceptance Ratio: 3.3%).



  • Shiyu Song, Manmohan Chandraker, Clark C. Guest, "High Accuracy Monocular SFM and Scale Correction for Autonomous Driving", published in IEEE Transactions on Pattern Analysis and Machine Intelligence (T-PAMI), Volume: 38, Issue: 4, April 1 2016, Page(s): 730 - 743, [link][project][pdf]

Shiyu Song, Manmohan Chandraker, "Robust Scale Estimation in Real-Time Monocular SFM for Autonomous Driving", In CVPR 2014: IEEE Conference on Computer Vision and Pattern Recognition, Columbus, Ohio, June 24-27, 2014, [link][project][pdf][video] [bibtex]. Acceptance Ratio: 29%.

Shiyu Song, Manmohan Chandraker, Clark C. Guest, "Parallel, Real-Time Monocular Visual Odometry", In ICRA 2013: IEEE International Conference on Robotics and Automation, Karlsruhe, Germany, May 6-10, 2013, [link] [project][pdf] [bibtex],


Professional Experience

  • Jan. 2016 - present, Senior Research Scientist (Tech Lead), Baidu Autonomous Driving Business Unit (Baidu ADU), Sunnyvale, CA USA
  • Nov. 2014 - Dec. 2015, Research Scientist, Baidu IDL, Sunnyvale, CA USA
  • Jan. 2014 - Oct. 2014, Senior Associate Researcher, NEC Laboratories America, Cupertino, CA USA
Patents
  • "Moving Object Localization in 3D Using a Single Camera", United States Patent # 9,070,202, grant on 06/31/2015, Assignee: NEC Labs America.
  • "Real-time Monocular Visual Odometry", United States Patent # 9,148,650,  grant on 09/29/2015, Assignee: NEC Labs America.
  • "Real-time Monocular Structure from Motion", United States Application # 13/858,041, filed on 04/06/2013, published on 05/22/2014, Applicant: NEC Labs America.