Joonkyu MinI am an incoming MS student at at Cognitive Learning for Vision and Robotics (CLVR) Lab at KAIST, where I will work on robot learning under supervision of Joseph J. Lim. I am currently an undergraduate student at Seoul National University, majoring in Electrical and Computer Engineering. Previously, I worked as a research intern at SNU VGI Lab under professor Jaesik Park, where I worked on 3D computer vision, especially 3D gaussian splatting. |
|
Research GoalMy research goal is to build safe and reliable physical intelligence. I am generally interested in robot learning and reinforcement learning. Particularly, I am interested in robot safety for reliable real-world robot training and deployment. One of my long-term goals is to automate scientific laboratories. By enabling robots to assist repetitive and labor-intensive lab tasks, I wish we can accelerate scientific discovery across fields such as biotechnology and life science. |
|
Publications |
|
CF3: Compact and Fast 3D Feature FieldsHyunjoon Lee, Joonkyu Min, Jaesik Park Int. Conf. on Computer Vision (ICCV), 2025 project page / paper / code / Proposed an approach for constructing a compact 3D feature representation from given 3D Gaussian Splatting |
ProjectsThese include coursework, side projects and unpublished research work. |
|
Can Zero-shot RL enable test time safety?project Robot Learning 2025-06-01 blog post / Exploring the potential of zero-shot RL method to enable safe decision-making at test time. |
|
KL-regularized FB representation for offline zero-shot RLproject Undergraduate 2025-05-30 code / Applied KL-divergence regularization approach to zero shot RL method (FB representation). |
PostsReinforcement Learning |
|
Design and source code from Jon Barron's website |