Sangjun Noh

I'm a Ph.D. candidate in the Gwangju Institute of Science and Technology (GIST), working in the GIST AILab under the supervision of Prof. Kyoobin Lee.

During my Ph.D., I have been primarily focused on generating large-scale datasets in physical simulators to advance robotic perception (e.g., unseen object understanding) and manipulation (e.g., grasping and placing). Recently, my research has expanded toward adapting Vision Foundation Models (VFMs) for robotics, learning policies, and exploring Vision-Language-Action (VLA) frameworks. Ultimately, my goal is to enable robots to perform human-like manipulation in the real world.

I genuinely enjoy discussing ideas and collaborating on exciting research. If you're interested in robotic perception, policy learning, or VLA, please feel free to reach out !! I’d love to connect! 😊

Email / CV / Scholar / Github / LinkedIn

🇰🇷 South Korea

Updates

[Apr '26]	Successfully defended Ph.D. thesis at GIST: "Scaling Up Robotic Manipulation via Foundation Priors and Large-Scale Synthetic Data".
[Jan '26]	Our work ManipForce was accepted to ICRA 2026!!
[Jan '26]	Our work BiGraspFormer was accepted to ICRA 2026!!
[Jun '25]	Our work GraspClutter6D was accepted to IEEE RA-L 2025!!
[May '25]	Our work GraspSAM was accepted to ICRA 2025!!
[May '24]	Our work DPLOT was accepted to CVPR 2024!!
[Mar '24]	Our work UOP was accepted to IEEE RA-L 2024!!
[Sep '22]	Started the Ph.D. Course at GIST AILab.
[Jul '22]	Successfully defended M.S. thesis at GIST.
[May '22]	Our work UOAIS was accepted to ICRA 2022!!
[Sep '20]	Started the M.S. program in Robotics & AI at GIST AI Lab.
[Jan '20]	Joined GIST AI Laboratory as a research intern.
[Aug '19]	Graduated from Dankook University (B.S. in Mechanical Engineering).

Research

The full list of my publications: Google Scholar

	3D Flow Diffusion Policy: Visuomotor Policy Learning via Generating Flow in 3D Space Sangjun Noh, Dongwoo Nam, Kangmin Kim, Geonhyup Lee, Yeonguk Yu, Raeyoung Kang, and Kyoobin Lee arXiv, 2025 paper / website Propose policy learning via generating scene flow in 3D space.
	ManipForce: Force-Guided Policy Learning with Frequency-Aware Representation for Contact-Rich Manipulation Geonhyup Lee, Youngjin Lee, Kangmin Kim, Seongju Lee, Sangjun Noh, Seunghyeok Back, and Kyoobin Lee IEEE International Conference on Robotics and Automation (ICRA), 2026 paper / website Propose vision & FT UMI for learning contact-rich manipulation
	BiGraspFormer: End-to-End Bimanual Grasp Transformer Kangmin Kim, Seunghyeok Back, Geonhyup Lee, Sangbeom Lee, Sangjun Noh, and Kyoobin Lee IEEE International Conference on Robotics and Automation (ICRA), 2026 paper / website Propose a transformer for bimanual grasping.
	GraspClutter6D: A Large-Scale Dataset for Robotic Grasping and Perception in Clutter Seunghyeok Back, Joosoon Lee, Kangmin Kim, Heeseon Rho, Geonhyup Lee, Raeyoung Kang, Sangbeom Lee, Sangjun Noh, Youngjun Lee, Taeyeop Lee, and Kyoobin Lee IEEE Robotics and Automation Letters (RA-L), 2025 paper / website / code Large-scale perception & 6d grasping benchmark in cluttered scenes.
	GraspSAM: When Segment Anything Model meets Grasp Detection Sangjun Noh, Jongwon Kim, Dongwoo Nam, Seunghyeok Back, Raeyoung Kang, and Kyoobin Lee IEEE International Conference on Robotics and Automation (ICRA), 2025 paper / website / code Extend SAM to the grasp detection.
	Domain-Specific Block Selection and Paired-View Pseudo-Labeling for Online Test-Time Adaptation Yeonguk Yu, Sungho Shin, Seunghyeok Back, Minhwan Ko, Sangjun Noh, and Kyoobin Lee CVPR, 2024 paper / code Online TTA with block selection + paired-view pseudo labels.
	Learning to Place Unseen Objects Stably Using a Large-Scale Simulation Sangjun Noh, Raeyoung Kang, Taewon Kim, Seunghyeok Back, Seongho Bak, and Kyoobin Lee IEEE Robotics and Automation Letters (RA-L)* — Gold Prize, 29th Samsung Humantech Paper Award paper / website / code [Sim2Real] Unseen object placement + stable placement.
	PolyFit: A Peg-in-hole Assembly Framework for Unseen Polygon Shapes via Sim-to-real Adaptation Geonhyup Lee, Joosoon Lee, Sangjun Noh, Minhwan Ko, Kangmin Kim, and Kyoobin Lee IROS, 2024 paper / website / [Sim2Real] Unseen object peg-in-hole.
	Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling Seunghyeok Back, Joosoon Lee, Taewon Kim, Sangjun Noh, Raeyoung Kang, Seongho Bak, and Kyoobin Lee ICRA, 2022 paper / website / code [Sim2Real] Unseen object amodal instance segmentation.

Awards & Honors

[2025]	Outstanding Paper Award, ICROS-KROS 2025
[2023]	Gold Prize (Top 0.7%), 29th Samsung Humantech Paper Award, Samsung Electronics
[2022]	Outstanding Paper Award (Best Paper in Mechanical Engineering), ICROS-KROS 2022
[2022]	Best Robot Vision Paper Award, Asian Federation of Computer Vision (AFCV), KROC 2022

Teaching

[Jul '25]	Teaching Assistant, AI & Robotics Summer School 2025, Korea Robotics Society (GIST) Hands-on deep learning (classification, detection, segmentation, generation, foundation models) · NVIDIA Isaac Sim for robotic grasping
[Jul '24]	Teaching Assistant, AI & Robotics Summer School 2024, Korea Robotics Society (GIST) Intro to Robotics AI (VLA) · NVIDIA Isaac Sim for robotic grasping
[Jul '23]	Teaching Assistant, AI & Robotics Summer School 2023, Korea Robotics Society (GIST) Intro to Robotics AI (Sim2Real) · NVIDIA Isaac Sim for robotic grasping

Imitation is the highest form of flattery