Papers
5
Total Citations
22
H-Index
2
About
Jae-hyeon Park is a robotics researcher whose work spans anomaly detection, surgical robotics, multi-object tracking, visual-inertial odometry, and LLM-based task planning. His most cited paper introduces a model-free unsupervised anomaly detection method using a stacked LSTM, designed to work on any robot controlled by feedback control—demonstrated on a fixed-wing UAV (11 citations). He has also contributed to surgical robotics by proposing an automated endoscopic camera manipulator for the da Vinci system using Rapidly-Exploring Random Trees (6 citations). In multi-object tracking, Park developed FocoTrack, which addresses the challenge of tracking at low frame rates by focusing on object overlap, a critical issue for resource-constrained robots. His work on visual-inertial odometry introduces a novel paradigm using stable embedding to handle manifold structures, enabling robust navigation in GPS-denied environments. Most recently, Park has explored autonomous task planning with large language models, tackling the challenge of executing abstract commands. With contributions that bridge theoretical foundations and practical robotic applications, Park’s research demonstrates a consistent focus on making robots safer, more autonomous, and more capable in real-world settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5