Sung Ju Hwang

Korea Advanced Institute of Science and Technology

Papers

Total Citations

H-Index

About

Sung Ju Hwang is a prominent researcher whose work spans computer vision, robotics, and machine learning, with particular expertise in object detection and meta-reinforcement learning. His research addresses some of the most pressing challenges in deploying intelligent systems in real-world environments. Among his most recognized contributions is his work on localization uncertainty estimation for anchor-free object detection, which has garnered over 30 citations across related publications. This research tackles a critical limitation in safety-critical systems — such as surgical robots and autonomous vehicles — by equipping object detectors with the ability to reason about uncertainty arising from sensor noise and incomplete data, a meaningful advancement for reliable deployment in unstable environments. Hwang has also made notable strides in robot learning through his work on skill-based meta-reinforcement learning, addressing the well-known sample inefficiency that plagues deep reinforcement learning methods. By enabling faster adaptation to complex, long-horizon tasks, this research brings autonomous robotic systems closer to practical feasibility. His dual focus on perception reliability and efficient robot learning reflects a coherent vision: building AI systems that are not only capable but trustworthy and deployable in high-stakes, real-world scenarios.