Dohyeong Kim
Papers
2
Total Citations
6
H-Index
2
About
Dohyeong Kim is a researcher whose work bridges the critical fields of satellite remote sensing and safe artificial intelligence. His early, foundational contribution involves the precise calibration of the COMS (Communication, Ocean, and Meteorological Satellite) visible channel using moon observation data. This work, published in 2018, is essential for ensuring the accuracy of long-term climate data records, directly supporting the Global Space-based Inter-Calibration System (GSICS) and the Korea Meteorological Administration (KMA) in maintaining data integrity from space-based platforms. More recently, Kim has made a significant leap into the domain of safe reinforcement learning (RL). In his 2023 work, he introduced a novel "trust region-based" approach for distributional RL that can simultaneously handle multiple safety constraints—a critical requirement for real-world robotic tasks like collision avoidance and energy management. This contribution addresses a core challenge in deploying RL in safety-critical environments, moving beyond single-constraint methods to ensure robust and reliable performance in complex systems. By tackling both the precision of Earth observation and the safety of autonomous decision-making, Kim demonstrates a rare versatility, applying rigorous mathematical frameworks to solve pressing problems in both geoscience and robotics.
Research Focus
Key Achievements
Top Papers
- 1COMS Visible Channel Calibration Using Moon Observation Data4 citations · 2018
- 2