Takuro Yamaguchi
Osaka University of Human Sciences, The University of Osaka, NTT (Japan)
Papers
4
Total Citations
49
H-Index
3
About
Takuro Yamaguchi is a pioneering researcher in multi-agent and real-world reinforcement learning, with a career spanning foundational work in cooperative AI to cutting-edge remote robotics. His early research, including the highly cited "Speed up reinforcement learning between two agents with adaptive mimetism" (21 citations), introduced innovative methods for selective knowledge sharing between agents, enabling faster learning without homogenizing behavior—a key challenge in multi-agent systems. Yamaguchi’s major contributions center on bridging the simulation-to-reality gap in robotics. His work "Realtime reinforcement learning for a real robot in the real environment" (18 citations) tackled the computational and temporal constraints of physical robot learning, while "Propagating learned behaviors from a virtual agent to a physical robot" (9 citations) advanced transfer learning from simulation to hardware. Most recently, Yamaguchi has achieved a notable breakthrough in remote robot control, co-developing a low-latency bilateral control system with haptic feedback (2024). This technology, a collaboration between NTT and Sony, enables operators to feel pressure and weight remotely, marking a significant leap in teleoperation. With a career that evolved from foundational reinforcement learning theory to practical, sensor-rich robotics, Yamaguchi’s work continues to influence both autonomous agents and human-robot interaction.
Research Focus
Key Achievements
Top Papers
- 1Speed up reinforcement learning between two agents with adaptive mimetism21 citations · 2002
- 2Realtime reinforcement learning for a real robot in the real environment18 citations · 2002
- 3
- 4