Seongho Bak

Gwangju Institute of Science and Technology

Papers

Total Citations

H-Index

About

Seongho Bak is a robotics researcher whose work lies at the intersection of computer vision and physical interaction, with a focus on enabling robots to perceive and manipulate objects in unstructured, cluttered environments. His most impactful contribution, the 2022 paper "Unseen Object Amodal Instance Segmentation via Hierarchical Occlusion Modeling" (66 citations), tackles a critical challenge: segmenting not just the visible portions of objects but their full, occluded shapes. This amodal perception capability is essential for robots to plan grasps and interactions in real-world scenes where objects are often stacked or partially hidden. Bak extends this theme of robust physical reasoning in his 2024 work on "Learning to Place Unseen Objects Stably Using a Large-Scale Simulation" (4 citations), which addresses the difficult task of determining stable placements for novel objects without requiring complete 3D models. Complementing his perception work, his 2020 paper on a "Soft Exosuit System for Walking Assistance During Stair Ascent and Descent" (26 citations) demonstrates a broader interest in wearable robotics and human-robot interaction. Through his research, Bak is building the perceptual and physical foundations for robots that can operate safely and effectively in the messy, unpredictable environments of homes and workplaces.