Geonho Cha

Seoul National University, Naver (South Korea)

Papers

Total Citations

H-Index

About

Geonho Cha is a robotics researcher whose work sits at the intersection of natural language processing, computer vision, and autonomous manipulation. His primary research areas include human-robot collaboration, 3D reconstruction, and novel sensing modalities for robotic perception. Cha’s most cited work, “Interactive Text2Pickup Networks for Natural Language-Based Human–Robot Collaboration” (2018, 32 citations), addresses a critical challenge in human-robot interaction: enabling robots to resolve ambiguous verbal commands during object pick-up tasks. This work introduced the IT2P network, a framework that allows robots to interactively clarify user intent, significantly improving the robustness of collaborative tasks. In a different vein, his research on “Deep Ego-Motion Classifiers for Compound Eye Cameras” (2019, 5 citations) explores bio-inspired vision systems, leveraging the unique hemispherical structure of insect eyes to achieve wide-field-of-view motion estimation with low aberration—a promising direction for compact robotic platforms. Most recently, his 2024 work “See-Then-Grasp” proposes a two-stage active reconstruction method that overcomes occlusion challenges in full 3D object modeling, enabling robots to reconstruct previously inaccessible regions like the bottom or back of objects. This contribution is particularly valuable for industrial and service robotics where complete object understanding is essential for reliable grasping. Cha’s research consistently demonstrates a commitment to bridging perception and action, making him a notable figure in the advancement of intelligent, interactive robotic systems.