Kwang Bin Lee

KAIST Graduate School of Culture and Technology.

me.jpg

Seoul, South Korea

Hey everyone, I’m Kwang Bin, currently a Master’s student at KAIST’s Graduate School of Culture Technology, advised by Prof. Sung-Hee Lee in the LAVA Lab. My research sits at the intersection of spatial intelligence and task planning for embodied agents, using foundation models to help them actually understand the 3D scenes they’re standing in and translate a task into scene-aware, visually grounded actionable steps — a skill I’ve always admired in humans, too.

I also really love books and films. My favorite author is Roald Dahl, and one of my favorite movies is Paddington!

news

Mar 05, 2026 :microphone: I will be presenting Situately as a Research Demo at IEEE VR 2026! :sparkles:
Mar 05, 2026 :sparkles: Our paper MaskAdapt has been accepted to CVPR 2026! :smile:

Research Works

  1. park2026maskadapt.jpg
    MaskAdapt: Learning Flexible Motion Adaptation via Mask-Invariant Prior for Physics-Based Characters
    Soomin Park, Eunseong Lee, Kwang Bin Lee, and Sung-Hee Lee
    In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2026
  2. FIGURE4_POLISHED.png
    Scene-agnostic Hierarchical Bimanual Task Planning via Visual Affordance Reasoning
    Kwang Bin Lee, Jiho Kang, and Sung-Hee Lee
    2025
    arxiv preprint
  3. ismar_demo.png
    Situated Embodied XR Agents via Spatial Reasoning and Prompting
    Jihun Shin, Hyeonjin Kim, Eunseong Lee, Donghwan Shin, Kwang Bin Lee, Taehei Kim, Hyeshim Kim, Joonsik An, and Sung-Hee Lee
    In 2025 IEEE International Symposium on Mixed and Augmented Reality Adjunct (ISMAR-Adjunct), 2025
    Research Demo