Dong Kyu Shin

Logo

M.S. Student
KAIST Graduate School of AI
Advised by Prof. Kimin Lee

View My GitHub Profile

Contact Info

About Me

Hi, I’m a master’s student at KAIST AI, advised by Prof. Kimin Lee. I received B.S. degrees in Mechanical Engineering and Artificial Intelligence from Seoul National University in August 2026. My research focuses on sequential decision-making, reinforcement learning, and robot learning from human data, with a particular interest in natural human–robot interaction.

Research Interests

Education

Korea Advanced Institute of Science and Technology (KAIST)
M.S. in Artificial Intelligence
Advisor: Kimin Lee
Sep. 2026 – Present
Seoul National University (SNU)
B.S. in Mechanical Engineering
B.S. in Interdisciplinary Major in Artificial Intelligence
Mar. 2019 – Aug. 2026
Shanghai Community International School
May 2015 – May 2018

Research & Professional Experience

KAIST AI · RISE Lab
Master's Student (Sep. 2026 – Present)
Research Intern (Aug. 2025 – Aug. 2026)
Aug. 2025 – Present

Advisor: Kimin Lee

  • Conducting research on Human-to-Robot Transfer and Human–Robot Interaction
  • Investigating how human observations can enhance robot learning capabilities
SNU Robotics Laboratory
Undergraduate Thesis
Aug. 2024 – Jun. 2025

Advisor: Frank Chongwoo Park

  • Developed SE(3)-equivariant grasp pose generation methods
  • Applied rectified flow models to 6-DoF grasp sampling
SNU Robotics Laboratory
Undergraduate Research Program
Jun. 2023 – Oct. 2023

Advisor: Frank Chongwoo Park

  • Studied anomaly detection algorithms
  • Implemented hyperspherical embedding with ViT backbone, achieving SOTA-comparable performance
Republic of Korea Army
Mandatory Military Service
Feb. 2021 – Jul. 2022
  • Discharged as Army Sergeant (Ammunition Management)

Publications

Publication Type — C: Conference | J: Journal | P: Preprint
(* equal contribution, † co-corresponding authors)

EDITH
[P1] EDITH: Hierarchical Policies from Verbal and Egocentric Human Signals for Natural Human–Robot Interaction
Dongjun Lee*, Juheon Choi*, Dong Kyu Shin*, Sinjae Kang, Kimin Lee
Under review / ArXiv Preprint, 2026
project page / arXiv

A hierarchical policy framework that learns natural human–robot interaction from verbal and egocentric human signals, enabling robots to interpret spoken instructions and first-person visual cues for grounded, intuitive collaboration.

XVR
[C2] Learning Multi-View Spatial Reasoning from Cross-View Relations
Suchae Jeong*, Jaehwi Song*, Haeone Lee, Hanna Kim, Jian Kim, Dongjun Lee, Dong Kyu Shin, Changyeon Kim, Dongyoon Hahm, Woogyeol Jin, Juheon Choi, Kimin Lee†
CVPR 2026
project page / arXiv / code

A large-scale dataset of 100K multi-view VQA samples that teaches VLMs spatial reasoning across viewpoints—correspondence, verification, and localization—and transfers those representations to robotic manipulation.

Value Gradient Sampler
[C1] Value Gradient Sampler: Learning Invariant Value Functions for Equivariant Diffusion Sampling
Himchan Hwang*, Hyeokju Jeong*, Dong Kyu Shin*, Che-Sang Park, Sehee Kweon, Sangwoong Yoon†, Frank Chongwoo Park†
AISTATS 2026
arXiv / code

A diffusion sampler parameterized by invariant value functions, inducing equivariant gradient flows without equivariant architectures and combining RL training with efficient invariant networks for high-quality particle sampling.

Selected Projects

RC-Car Autonomous Racing Challenge
2024
video

Modified PPO-based reinforcement learning for autonomous racing on unknown tracks, using Python, ROS2, and MuJoCo.

Technical Skills

Programming Languages: Python (PyTorch, OpenGL), MATLAB, C++, ROS2
Languages: Korean (Native), English (Intermediate), Mandarin Chinese (Conversational), French (Beginner)

Awards and Honors

Hobbies

Saxophone

I enjoy playing the saxophone in my free time. Learning and performing improvisation is one of the most creatively fulfilling parts of my life.

View Performances
Playing saxophone

Last updated: September 2026