Keisuke Nakarnura

Papers

Total Citations

H-Index

About

Keisuke Nakamura is a prominent researcher in reinforcement learning (RL) and robot control, with a focus on enabling autonomous systems to operate in complex, uncertain environments. His work addresses a fundamental challenge in RL: the difficulty of designing efficient reward functions for high-dimensional tasks. In his highly cited 2023 paper, "Model-based Adversarial Imitation Learning from Demonstrations and Human Reward," Nakamura pioneered a novel approach that combines model-based RL with adversarial imitation learning and human feedback. This method allows robots to learn complex behaviors directly from demonstrations and sparse human reward signals, bypassing the need for hand-crafted reward functions. With 7 citations in its first year, this work has already influenced the field of imitation learning and human-in-the-loop robotics. Nakamura's contributions are particularly impactful for real-world applications, where reward engineering is often impractical. His research bridges the gap between theoretical RL and practical robot control, offering scalable solutions for tasks ranging from manipulation to navigation. As a rising figure in the RL community, Nakamura continues to push the boundaries of how robots can learn from both data and human guidance.