Teyun Kwon

Papers

Total Citations

H-Index

About

Teyun Kwon is a rising researcher at the intersection of robotics and artificial intelligence, with a primary focus on leveraging large language models (LLMs) for robotic control and planning. His most cited work, "Language Models as Zero-Shot Trajectory Generators" (2024, 46 citations), challenges the prevailing assumption that LLMs are only useful for high-level task planning. Kwon demonstrates that LLMs can directly generate low-level robot trajectories without any fine-tuning or task-specific training, effectively acting as zero-shot motion planners. This contribution bridges a critical gap between language understanding and physical action, suggesting that pre-trained language models already encode enough spatial and kinematic knowledge to guide a robot's movements. By showing that an LLM can output joint angles or end-effector positions from a natural language command, Kwon opens new avenues for more intuitive human-robot interaction and reduces the need for extensive reward engineering or imitation learning datasets. His work is particularly notable for its simplicity and elegance—using nothing more than prompt engineering to unlock a capability previously thought to require specialized models. As a young researcher, Kwon's findings have already sparked follow-up studies in embodied AI and are influencing how the robotics community views the role of foundation models in low-level control.