Papers
5
Total Citations
81
H-Index
3
About
Seunghyeok Back is a leading researcher in robotic perception and manipulation, whose work focuses on enabling robots to interact intelligently with unseen objects in unstructured, cluttered environments. His major contributions span amodal instance segmentation, grasp detection, and stable object placement. Back’s most influential paper, “Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling” (2022, 66 citations), broke new ground by moving beyond visible-region segmentation to infer the full shape of objects even when partially occluded—a critical capability for real-world robotic grasping. He further advanced the field with GraspSAM (2025), an innovative extension of the Segment Anything Model that enables prompt-driven, user-guided grasp detection without prior object knowledge. His work on learning to place unseen objects stably using large-scale simulation (2024) addresses the fundamental challenge of object placement for partially observed items. Back also contributed to understanding assembly instructions through context-aware data augmentation (2021) and introduced GraspClutter6D (2025), a large-scale real-world dataset designed to push robust perception and grasping beyond simplistic benchmarks. Through these contributions, Back is shaping the future of autonomous robotic manipulation in complex, real-world settings.
Research Focus
Key Achievements
Top Papers
- 1
- 2GraspSAM: When Segment Anything Model Meets Grasp Detection6 citations · 2025
- 3Learning to Place Unseen Objects Stably Using a Large-Scale Simulation4 citations · 2024
- 4
- 5