Songhwai Oh
Papers
44
Total Citations
491
H-Index
12
About
Songhwai Oh is a prolific robotics and artificial intelligence researcher whose work spans motion planning, safe reinforcement learning, human-robot collaboration, and multi-agent systems. His research has made significant contributions to enabling robots to operate intelligently in complex, real-world environments — from navigating crowded pedestrian spaces to collaborating naturally with humans through language. Among his most influential contributions is a cross-entropy-based cost-aware path planning framework (2017, 53 citations) that dramatically improves upon classical sampling-based methods like RRT, alongside a Gaussian process motion controller (2014, 42 citations) enabling real-time navigation through dynamic crowd scenarios. His vision-based coordinated localization work (2014, 37 citations) addresses critical challenges for GPS-denied mobile sensor networks. Oh has also advanced the frontier of safe reinforcement learning through his Trust Region Conditional Value at Risk (TRC) framework, providing principled risk-aware policy optimization crucial for real-world robotics deployment. His interest in unconventional robotic platforms is demonstrated through entropy-adaptive reinforcement learning applied to soft vibration-actuated tripod robots (2020, 27 citations). Collectively accumulating hundreds of citations, Oh's body of work reflects a consistent vision of building robots that are safe, adaptive, and capable of meaningful human collaboration.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3Vision-Based Coordinated Localization for Mobile Sensor Networks37 citations · 2014
- 4
- 5
- 6
- 7TRC: Trust Region Conditional Value at Risk for Safe Reinforcement Learning21 citations · 2022
- 8
- 9Toward Robotic Sensor Webs: Algorithms, Systems, and Experiments20 citations · 2011
- 10