Weikun Peng
Weikun Peng

CS PhD Student

I am a first-year PhD student at Simon Fraser University, advised by Prof. Manolis Savva. I received Master degree at National University of Singapore, where I spent wonderful two years working on robotic manipulation with Prof. Lin Shao. I received Bachelor degree at Beihang University. In addition, I was also fortunate to collaborate with Prof. Cewu Lu at Shanghai Jiao Tong University and Dr. Yan Wang at SenseTime Research. My research interests lie in the intersection of computer graphics and robotics. Recently, I’m curious about modeling the functionality of objects from ego-centric videos and thus pushing forward the Real-to-Sim-to-Real learning paradigm in robotics.

Outside of research, I enjoy hiking, cooking and reading. My favorite novelist is Stefan Zweig.

Interests
  • • Computer Graphics
  • • Robotics
  • • Literature
Education
  • Ph.D. in Computer Science

    Jan 2025 - present

    Simon Fraser University University

  • Msc. in Artificial Intelligence

    Aug 2022 - Jun 2024

    National University of Singapore

  • B.Eng. in Computer Science

    Sep 2018 - Jun 2022

    Beihang University

Publications
Generalizable Articulated Object Reconstruction from Casually Captured RGBD Videos
Generalizable Articulated Object Reconstruction from Casually Captured RGBD Videos
We develop a method that can reconstruct articulated objects from casually captured RGBD videos.
TieBot: Learning to Knot a Tie from Visual Demonstration through a Real-to-Sim-to-Real Approach
TieBot: Learning to Knot a Tie from Visual Demonstration through a Real-to-Sim-to-Real Approach
We develop a Real-to-Sim-to-Real approach that enables learning tie-knotting skills for robots.
ManiFoundation Model for General-Purpose Robotic Manipulation of Contact Synthesis with Arbitrary Objects and Robots
ManiFoundation Model for General-Purpose Robotic Manipulation of Contact Synthesis with Arbitrary Objects and Robots
We develop a foundation model on robotic manipulation via modeling contact points on the object.
ELIC: Efficient Learned Image Compression with Unevenly Grouped Space-Channel Contextual Adaptive Coding
ELIC: Efficient Learned Image Compression with Unevenly Grouped Space-Channel Contextual Adaptive Coding
We develop a deep image compression model that leverages the uneven information distribution within the latent variables for efficient image compression.

Experience

  1. logo Research intern at MVIG group

    Shanghai Jiao Tong University
    I worked with Jun Lv on robotic manipulation and articulated object reconstruction projects.
  2. logo Research intern at SenseTime Research

    SenseTime
    I worked with Dailan He and Dr. Yan Wang on deep image compression and image compression for machine perception.
Side Projects
Post