XIANG LI

xiangli8 \at cs \dot stonybrook \dot edu

I am a Ph.D. candidate in CS working with Prof. Michael S. Ryoo at SBU.
My research interest includes computer vision and robotics.
I was a master student with Prof. Yue Zhou at SJTU.

Education

Stony Brook University

Ph.D. Program in Computer Science
Jan. 2020 - Present

Shanghai Jiao Tong University

Master Programe of Control Engineering
Pattern Recognition and Intelligent System
Sept. 2015 - March 2018

Shanghai Jiao Tong University

Undergraduate Programe of Automation
Sept. 2011 - June 2015

Selected Publication

  • Neural Neural Textures Make Sim2Real Consistent [Paper][Code]

    Ryan Burgert, Jinghuan Shang, Xiang Li and Michael S. Ryoo, June 2022
  • Does Self-supervised Learning Really Improve Reinforcement Learning from Pixels? [Paper]

    Xiang Li, Jinghuan Shang, Srijan Das and Michael S. Ryoo, June 2022
  • StARformer: Transformer with State-Action-Reward Representations [Paper][Code]

    ECCV'22
    Jinghuan Shang, Kumara Kahatapitiya, Xiang Li and Michael S. Ryoo, Oct. 2021

Selected Projects

  • Simple Raspi Car [Code]

    Arduino C++ Python

    An open-source solution to transform general RC car into powerful mobile platform with Raspberry Pi 3B+ even NVIDIA TX2. It is capable for carrying heavier payload hardware, solving some larger A.I. problem and easy to maintain or upgrade.
  • Project PI Car

    C++ Python

    Project PI Car aimed to build a simple mobile platform, based on Raspberry PI 3B, for many interesting A.I. tasks. The car contains one encoder, two cameras, and a Raspberry PI 3B as controller. I designed the layout of the car and two SDKs for both C++ and Python users. We have produced 29 cars so far and keep on making improvements.
  • qtCyberDIP [Code][Video]

    C++

    qtCyberDIP is a windows driver for a device called CyberDIP. The device (China Utility Model Patent No.2016201772460), which is also designed by me, can touch phone/pad's screen like a human finger. Computer can play mobile games autonomously with the information from screen capture. We used it to perform several image processing experiments, and I am the TA of Digial Image Processing responsible for these experiments. Check code and video for further information.
  • Tracking with cradle camera

    C++

    I combined CT and STC with a cradle camera to tracking moving vehicles. The cradle was controlled by a PI controller via PELOC-D. GUI was based on MFC.
  • FlyCapture Bundle [Code]

    C++

    A toolkit to drive Point-Grey USB camera. Based on SDK, I added an auxiliary focusing and auto-exposure module with several different modes. Multi-thread and buffering feature were included to keep all cameras synchronized.

Awards

  • Outstanding graduates in Shanghai, 2018
  • National scholarship for graduate students, 2017
  • Outstanding graduates of SJTU, 2015

Last updated at July 13, 2022 | Back to top