Avatar

宋适宇

Director of Localization and Mapping

TuSimple USA

Biography

I joined TuSimple USA in June, 2021. We are building state-of-the-art localization, calibration, and mapping technologies for self-driving trucks.

I was one of the founding team members of the Baidu Autonomous Driving Car project. I joined Baidu in 2014. Prior to joining Baidu, I was a research scientist in the media analysis group in NEC Labs America, Cupertino CA.

My research interests include Computer Vision, Robotics, Machine Learning, Simultaneous Localization and Mapping (SLAM).

Interests

  • Computer Vision
  • Machine Learning
  • Simultaneous Localization and Mapping (SLAM)
  • Autonomous Driving

Education

  • Ph.D. in Electrical and Computer Engineering

    University of California, San Diego

  • M.S. in Electrical and Computer Engineering

    University of California, San Diego

  • BSc in Electrical Engineering

    Tsinghua University

Accomplish­ments

The 20th China Patent Award - Silver Award

Issuer: China National Intellectual Property Administration (CNIPA) Dec 2018
Patent: ZL201610348334.7 无人车、无人车定位方法、装置和系统，百度在线网络技术（北京）有限公司，宋适宇、李文博、张天雷

Featured Publications

Diff-Net: Image Feature Difference based High-Definition Map Change Detection

Up-to-date High-Definition (HD) maps are essential for self-driving cars. To achieve constantly updated HD maps, we present a deep …
PDF Video

Exploring Imitation Learning for Autonomous Driving with Feedback Synthesizer and Differentiable Rasterization

We present a learning-based planner that aims to robustly drive a vehicle by mimicking human drivers’ driving behavior. We …
PDF Video

DA4AD: End-to-end Deep Attention-based Visual Localization for Autonomous Driving

We present a visual localization framework based on novel deep attention aware features for autonomous driving that achieves centimeter …
PDF Dataset Project Video

DeepVCP: An End-to-End Deep Neural Network for Point Cloud Registration

We present DeepVCP - a novel end-to-end learning-based 3D point cloud registration framework that achieves comparable registration …
PDF Dataset Video DOI

Recent Publications

Quickly discover relevant content by filtering publications.
Lei He, Shengjie Jiang, Xiaoqing Liang, Ning Wang, Shiyu Song . Diff-Net: Image Feature Difference based High-Definition Map Change Detection. In ICRA, 2022.

PDF Video

Jinyun Zhou, Rui Wang, Xu Liu, Yifei Jiang, Shu Jiang, Jiaming Tao, Jinghao Miao, Shiyu Song . Exploring Imitation Learning for Autonomous Driving with Feedback Synthesizer and Differentiable Rasterization. In IROS, 2021.

PDF Video

Yao Zhou, Guowei Wan, Shenhua Hou, Li Yu, Gang Wang, Xiaofei Rui, Shiyu Song . DA4AD: End-to-end Deep Attention-based Visual Localization for Autonomous Driving. In ECCV, 2020.

PDF Dataset Project Video

Wendong Ding, Shenhua Hou, Hang Gao, Guowei Wan, Shiyu Song . LiDAR Inertial Odometry Aided Robust LiDAR Localization System in Changing City Scenes. In ICRA, 2020.

PDF Project Video

Weixin Lu, Guowei Wan, Yao Zhou, Xiangyu Fu, Pengfei Yuan, Shiyu Song . DeepVCP: An End-to-End Deep Neural Network for Point Cloud Registration. In ICCV, 2019.

PDF Dataset Video DOI

Weixin Lu, Yao Zhou, Guowei Wan, Shenhua Hou, Shiyu Song . L3-Net: Towards Learning based LiDAR Localization for Autonomous Driving. In CVPR, 2019.

PDF Dataset Project Video DOI

Guowei Wan, Xiaolong Yang, Renlan Cai, Hao Li, Yao Zhou, Hao Wang, Shiyu Song . Robust and Precise Vehicle Localization based on Multi-sensor Fusion in Diverse City Scenes. In ICRA, 2018.

PDF Project Video DOI

Shiyu Song, Manmohan Chandraker, Clark C. Guest . High Accuracy Monocular SFM and Scale Correction for Autonomous Driving. In T-PAMI, 2016.

PDF Project DOI

Shiyu Song, Manmohan Chandraker . Joint SFM and Detection Cues for Monocular 3D Localization in Road Scenes. In CVPR (Oral), 2015.

PDF Project Poster Video DOI

Shiyu Song, Manmohan Chandraker . Robust Scale Estimation in Real-Time Monocular SFM for Autonomous Driving. In CVPR, 2014.

PDF Project Poster Video DOI

Shiyu Song, Manmohan Chandraker, Clark C. Guest . Parallel, Real-Time Monocular Visual Odometry. In ICRA, 2013.

PDF Project DOI

Recent & Upcoming Talks

Localization and Mapping Technology Under the Hood of World First "Driver out" Fully Autonomous Semi-Truck

TuSimple's AI-based, Level 4 autonomous trucking technology addresses today's greatest industry challenges. Recently, TuSimple …
Jun 5, 2022 9:00 AM IV2022, Aachen, Germany

Towards Learning-based Localization for Autonomous Driving (Chinese)

We make the introduction of several key techniques in our self-driving car, including a multi-sensor fusion based localization system, …
Sep 15, 2020 2:00 PM Baidu World Conference 2020, Beijing, China
Video

Inside Apollo: Multisensor Fusion Based Localization

We make a brief introduction of the techniques behind our multi-sensor fusion based localization system. We present a robust and …
Jun 17, 2019 9:00 AM CVPR 2019, Long Beach, California, U.S.A.
Slides

Towards Learning-based Localization for Autonomous Driving

We present a robust and precise localization system that achieves centimeter-level localization accuracy in varied city scenes. Our …
Mar 28, 2019 9:00 AM MIPR 2019, San Jose California, U.S.A.

Localization and HD Map at Baidu IDG

We make brief introduction of HD Map services at Baidu Apollo platform and the techniques behind our multi-sensor fusion based …
Mar 28, 2018 9:00 AM Baidu AI Developer Conference 2018, Beijing, China

HD Map, Localization and Self-Driving Car

We make brief introduction of the development history of Baidu self-driving car, Baidu Apollo platform for developers, the techniques …
Sep 19, 2017 9:00 AM The ISPRS Geospatial Week, Wuhan Hubei, China

Projects

Multi-sensor Fusion based Localization System

A robust and precise vehicle localization system that achieves centimeter-level accuracy by adaptively fusing information from multiple …

Monocular Visual Odometry

A real-time monocular visual odometry system that corrects for scale drift using a novel cue combination framework for ground plane …

Contact