Max Robotics — Global Service Robot Directory + US Certification

About

Yoichiro Kawamura is a robotics researcher whose work sits at the intersection of imitation learning, adaptive control, and humanoid manipulation. His primary research areas include motion style transfer, environment-adaptive robot control, and skill acquisition through deep reinforcement learning. Kawamura’s major contributions involve integrating parametric bias into neural network architectures to enable robots to not only replicate human demonstrations but also adapt their motion style and behavior to varying environmental conditions. His 2021 paper on imitation learning with additional motion style constraints (15 citations) is his most cited work, introducing a method that allows robots to generalize tasks while preserving nuanced motion trajectories. In parallel, his research on stochastic predictive networks with parametric bias (5 citations) advances environmentally adaptive control by incorporating variance minimization, a critical capability for mobile robots operating in unpredictable settings. Kawamura has also tackled the challenging domain of tool force adjustment in life-sized humanoids (3 citations), using deep reinforcement learning combined with active teaching requests to overcome the difficulties of physical modeling. His work represents a meaningful step toward more flexible, human-like robotic behavior in real-world applications.