The KAIST Robotics and Vision Lab, led by Prof. Sunghoon Im, advances Physical AI for machines that perceive, reason, and act in the real world. Our research spans computer vision, robot learning, and multimodal generative AI — from 3D perception and scene understanding to vision-language-action models for robots and autonomous systems.
🚀 We are hiring! Join Us →
Recent achievements, publications, and lab activities
ReAL: Reference-to-Image (R2I) Aware Latent Diffusion for Image Super-Resolution Byeonghun Lee, Hyunmin Cho, Sunghoon Im†, and Kyong Hwan Jin† European Conference on Computer Vision ( ECCV ), 2026
Learning to Forget: Emotional Salience as a Compression Mechanism for Long-Term AI Memory SoYeop Yoo, Sunghoon Im† ACM SIGKDD Conference on Knowledge DIscovery and Data Minig ( KDD ), 2026
Our lab has been selected for the Core Research (Type C) Project. Title: Adaptive Policy Learning Technology for Collaborative Robots Understanding Physical Tacit Knowledge Duration: March 1, 2026 – February 28, 2029 (3 Years) Support: National Research Foundation of Korea (NRF)
Our project has been selected for 2026 Infrastructure Supporting Grants for Young Scientist Program. Through this program, we are planning to purchase Unitree H1‑2 humanoid robot to support our robotics and embodied AI research. Unitree H1-2 robot
Our project(Chain-of-Thought-based Video Generation Model following Physical Laws) has been selected for the Nvidia Academic Grant, providing access to 24 NVIDIA B200 (three NVIDIA B200 x 8) GPU server. https://www.nvidia.com/ko-kr/data-center/dgx-b200/ NVIDIA B200 GPU
Minjae Kim , Woosik Kim , Jaehoon Lee , Jiwon Jang joined our group as M.S. course. welcome all!
