Junichiro Yoshimoto

About

Junichiro Yoshimoto is a leading researcher in reinforcement learning and robotics, with a particular focus on applying advanced machine learning techniques to complex, nonlinear control problems. His work centers on the development of algorithms that enable robots to learn and perform challenging dynamic tasks, especially those involving continuous state and action spaces. Yoshimoto’s major contributions include pioneering the use of reinforcement learning for the balancing of the Acrobot—a notoriously difficult two-link robot with only a single actuator. His 2003 paper on this topic, which has garnered 19 citations, demonstrates how RL can effectively manage the Acrobot’s nonlinear dynamics, a milestone that has influenced subsequent research in robot control and autonomous learning. Earlier, in 2001, he explored the application of an on-line Expectation-Maximization (EM) algorithm to the same problem, laying foundational work for efficient, real-time learning in robotics. Through these studies, Yoshimoto has advanced the practical implementation of RL in continuous control tasks, bridging the gap between theoretical algorithms and real-world robotic systems. His research remains a key reference for students and engineers working on reinforcement learning, robot manipulation, and adaptive control.

Research Focus

Key Achievements

Top Papers

1. 1
   Application of reinforcement learning to balancing of Acrobot

   19 citations · 2003
2. 2
   Application of reinforcement learning based on on‐line EM algorithm to balancing of acrobot

   3 citations · 2001

Key Collaborators

Contact & Links