Yohei Hayamizu

University of Electro-Communications, Binghamton University

Papers

Total Citations

H-Index

About

Yohei Hayamizu is a robotics researcher whose work sits at the intersection of reinforcement learning, automated planning, and vision-language models for robot control. His research focuses on enabling robots to reason and act intelligently in complex, open-world environments. Hayamizu’s most influential work, “Guiding Robot Exploration in Reinforcement Learning via Automated Planning” (2021, 8 citations), introduces a novel framework that leverages symbolic planning to direct an RL agent’s exploration, significantly improving sample efficiency and task completion. This contribution bridges two traditionally separate fields—planning and learning—offering a principled way to combine high-level reasoning with low-level control. In his more recent work, “Learning Quadruped Locomotion Policies Using Logical Rules” (2024, 3 citations), he explores how logical specifications can generate diverse and robust locomotion gaits without requiring motion priors or extensive manual tuning. Additionally, with “DKPROMPT: Domain Knowledge Prompting Vision-Language Models for Open-World Planning” (2024, 3 citations), Hayamizu addresses the limitations of VLMs in robot planning by injecting domain-specific knowledge directly into prompts, enhancing their reasoning capabilities. His research is notable for its elegant integration of symbolic AI with modern learning-based methods, offering practical pathways toward more autonomous and adaptable robots.