Byeonghwi Kim

Seoul National University

Papers

2

Total Citations

6

H-Index

2

About

Byeonghwi Kim is an emerging researcher specializing in embodied AI, robotic instruction following, and multimodal learning — fields at the intersection of computer vision, natural language processing, and autonomous agents. His work focuses on enabling robotic systems to understand and execute complex tasks from natural language instructions, a challenging frontier in artificial intelligence research. Among his notable contributions, Kim developed ReALFRED, a benchmark designed to evaluate embodied instruction-following agents in photo-realistic environments, pushing the field toward more realistic and rigorous evaluation standards. His research on multi-modal grounded planning tackles a critical bottleneck in training robotic assistants: the high cost of large-scale language annotations. By leveraging large language models to reduce annotation overhead, his work makes embodied agent training significantly more accessible and scalable. Though early in its citation trajectory — with his recent works accumulating citations in 2024–2025 — Kim's research addresses fundamental challenges that will shape the next generation of intelligent robotic assistants. His contributions are particularly relevant for researchers working on household robots, human-robot interaction, and practical deployment of AI agents in real-world environments. Students entering embodied AI will find his benchmark and planning frameworks valuable reference points.

Research Focus

Key Achievements

2
H-Index
2
Papers
6
Total Citations
3
Avg Citations/Paper
🏆 Most Cited Paper
ReALFRED: An Embodied Instruction Following Benchmark in Photo-Realistic Environments
3 citations · 2024
📈 Most Prolific Year: 2024 (1 Papers)
🤝 Key Collaborators: 5
🏛 Institutions: Seoul National University

Top Papers

  1. 1
  2. 2

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago