Taewhan Kim 김태환

Research Engineer · Neuromeka

I am a research engineer at Neuromeka, mentored by Dr. Joonho Lee, working on learning-based robot manipulation for real-world manufacturing. I am also fortunate to collaborate with Prof. Sehoon Ha at Georgia Tech, where I have been a visiting researcher since September 2026.

I received my M.S. in Computer Science from Peking University, advised by Prof. Hao Dong, and my B.Eng. in Electrical and Electronic Engineering from the University of Nottingham Ningbo China.

My research interests include contact-rich manipulation and deploying robot learning in industry.

Taewhan Kim

News

Publications

* equal contribution  ·  † corresponding author

Learning-Augmented Robotic Automation for Real-World Manufacturing
Yunho Kim, Quan Nguyen, Taewhan Kim, Youngjin Heo, Joonho Lee†
arXiv, 2026
arXiv Project Page Video

A production-grade bimanual cobot doing motor cable pick, insertion and soldering in a shared factory workspace. It reached 99.4% insertion reliability and assembled 108 motors over 5 hours.

ManipGPT: Is Affordance Segmentation by Large Vision Models Enough for Articulated Object Manipulation?
Taewhan Kim, Hojin Bae, Zeming Li, Xiaoqi Li, Iaroslav Ponomarenko, Ruihai Wu, Hao Dong†
IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2025
Project Page arXiv

Uses a large vision model to segment affordances so a robot knows where to interact with articulated objects, without complex datasets or perception pipelines.

CheckManual: A New Challenge and Benchmark for Manual-based Appliance Manipulation
Yuxing Long, Jiyao Zhang, Mingjie Pan, Tianshu Wu, Taewhan Kim, Hao Dong†
IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2025
Project Page Paper Code

A benchmark where robots must read an appliance's manual to operate it correctly.

RAD: A Realistic Multi-View Benchmark for Pose-Agnostic Anomaly Detection
Kaichen Zhou*, Xinhai Chang*, Taewhan Kim*, Jiadong Zhang*, Yang Cao, Chufei Peng, Fangneng Zhan, Hao Zhao, Hao Dong, Kai Ming Ting, Ye Zhu†
arXiv, 2024
Project Page arXiv Code

A real-world multi-view dataset for detecting anomalies in objects seen from any pose.

Learning Part-Aware Visual Actionable Affordance for 3D Articulated Object Manipulation
Yuanchen Ju*, Haoran Geng*, Ming Yang*, Yiran Geng, Yaroslav Ponomarenko, Taewhan Kim, He Wang, Hao Dong†
CVPR Workshop on 3D Vision and Robotics, 2023

Learns part-aware affordances that tell a robot which part of a 3D articulated object to act on.

Academic Service

Education & Experience

2026 – Present
Georgia Institute of Technology, Atlanta, GA, USA
Visiting Researcher · Advisor: Prof. Sehoon Ha
2025 – Present
Neuromeka, Seoul, Korea
Research Engineer · Advisor: Dr. Joonho Lee
2024 – 2025
AgiBot, Beijing, China
Research Intern
2022 – 2025
Peking University, Beijing, China
M.S. in Computer Science · Advisor: Prof. Hao Dong
2018 – 2022
University of Nottingham Ningbo China, Ningbo, China
B.Eng. in Electrical and Electronic Engineering