Ying-Sheng Luo
I use Computer Graphics for
About
Ying-Sheng Luo 羅應陞
I'm currently a research engineer from Inventec Corp., a world-leading computer and electronics manufacturer with more than 16 billion USD annual revenue. At Inventec AI Center, I focus on improving its manufacturing processes and pushing its technological advancements through deep learning algorithms.
- Phone: +886-912-378590
- Email: tray307969@gmail.com
- City: Taipei, Taiwan
- Language: English, Chinese (Native)
I received my M.Sc in Computer Science from the National Taiwan University of Science and Technology (2018), advised by Professor Yu-Chi Lai. My work lies in the intersection between computer graphics and machine learning, with a focus on reinforcement learning for motion control of simulated characters and real-time rendering.
I'm also a computer graphics enthusiast and like to apply computer graphics knowledge I've learned in my other hobbies & side projects. There are examples from my piano cover videos where I seamlessly added 3D virtual objects with convincing lighting and reflections to make the video more interesting.
Resume [PDF]
Education
Master of Computer Science
Sep 2016 - June 2018
National Taiwan University of Science and Technology, Taipei, Taiwan
- Thesis: Intuitive and Easily-Installed Fenestration and Light Control Framework
- Implemented precomputed radiance transfer method to estimate global illumination
- Graduated with 4.18 out of 4.3 GPA
Bachelor of Computer Science
Sep 2013 - June 2016
National Taiwan University of Science and Technology, Taipei, Taiwan
- Ranked 1 in undergrde class
- Received presidential award
- Graduated with 3.94 out of 4.0 GPA
Work Experience
Research Engineer
July 2018 - Present
Inventec Corp., Taiwan
- Deployed the RL policies trained in simulation to the real-world Unitree A1 quadruped robot to perform many skills using Sim2Real techniques
- Invent the transition motion tensor to enable controllable and versatile character movements in physics-enabled simulation
- Design a controller for simulated characters using Deep-RL and GAN such that the character produces natural movements while obeying high-level user directives
- Research and deploy state-of-the-art deep learning solutions for internal business units; 4 issued and pending patents
- Build an iOS App and backend Restful API server with Kafka job queues to deploy the AI medical product. The medical iOS App passed the security certificate licensed by Gapertise
Graphics Engineer Intern
June 2015 - Sep 2015
International Games System Corp., Taiwan
- Developed VFX effects using GLSL in a mobile car racing game
- Implemented color grading, dynamic clouds, lightmaps, and HDR cubemap using GLSL and Cocos2d
- Explored GPU performance analysis tool to improve rendering performance on mobile devices
Publications
Benchmarking Smoothness and Reducing High-Frequency Oscillations in Continuous Control Policies
IEEE International Conference on Intelligent Robots and Systems (IROS 2024)
(*Joint first authors)
[Paper] [Supplementary Video]
Expert Composer Policy: Scalable Skill Repertoire for Quadruped Robots
IEEE International Conference on Robotics and Automation (ICRA 2024)
(*Joint first authors)
Expanding Versatility of Agile Locomotion through Policy Transitions Using Latent State Representation
IEEE International Conference on Robotics and Automation (ICRA 2023)
(*Joint first authors)
Transition Motion Tensor: A Data-Driven Approach for Versatile and Controllable Agents in Physically Simulated Environments
SIGGRAPH Asia 2021 - Technical Communications
(*Joint first authors)
[Paper] [Code] [Supplementary Video]
CARL: Controllable Agent with Reinforcement Learning for Quadruped Locomotion
ACM Transactions on Graphics (SIGGRAPH 2020)
(*Joint first authors)
[Paper] [Code] [Supplementary Video] [Two Minute Papers Video] [SIGGRAPH 2020 Paper Preview]
Piano Videos
One of my hobby is playing the piano where I stared learning since 2019. Here are some of my piano videos.