Papers
11
Total Citations
120
H-Index
5
About
Hyeonwoo Yu is a robotics and computer vision researcher whose work sits at the intersection of simultaneous localization and mapping (SLAM), probabilistic 3D scene understanding, and neural rendering. Over nearly a decade of research, Yu has made sustained contributions to semantic SLAM, developing principled Bayesian and variational frameworks for encoding and reasoning about complex 3D objects within robot perception pipelines. His early papers on variational feature encoding and probabilistic object observation models laid important groundwork for high-level semantic mapping, accumulating nearly 40 citations combined and establishing him as a thoughtful contributor to probabilistic robotics. More recently, Yu has embraced cutting-edge neural scene representation techniques, including Neural Radiance Fields and 3D Gaussian Splatting. His 2024 paper "RGBD GS-ICP SLAM" has already garnered 52 citations, signaling strong community interest in his approach to dense visual SLAM using Gaussian splatting. Additional work on street-level localization for autonomous vehicles and multi-modal variational autoencoders for human-robot teaming reflects the breadth of Yu's vision. Across his career, Yu has consistently sought to bridge probabilistic inference with modern deep learning, making his research particularly relevant to students working in autonomous systems and 3D scene reconstruction.
Research Focus
Key Achievements
Top Papers
- 1RGBD GS-ICP SLAM52 citations · 2024
- 2A Variational Observation Model of 3D Object for Probabilistic Semantic SLAM25 citations · 2019
- 3
- 4
- 5
- 6
- 7A Bayesian approach to terrain map inference based on vibration features3 citations · 2017
- 8
- 9
- 10