Takumi Aotani

Meiji University, Nara Institute of Science and Technology

Papers

Total Citations

H-Index

About

Takumi Aotani is a researcher advancing the frontiers of reinforcement learning (RL) and multi-agent systems, with a focus on creating more expressive, efficient, and biologically inspired algorithms. His work addresses critical limitations in robotic control and cooperative AI. Aotani’s key contributions include the design of a restricted normalizing flow for stochastic policies, which enhances computational efficiency and expressiveness in RL—a breakthrough for acquiring complex behaviors. He also introduced the Reward-Punishment Actor-Critic (RP-AC) algorithm, a biologically grounded framework that improves robotic non-grasping manipulation, enabling robots to handle intricate tasks more safely and effectively. In multi-agent settings, Aotani developed a bottom-up approach for selective cooperation, tackling the inherent complexity of systems like cooperative transport without exhaustive preprogramming. His most cited papers, each garnering 3–4 citations, reflect a growing impact in the RL community. Aotani’s work is notable for bridging theoretical innovation with practical robotic applications, offering scalable solutions for autonomous systems. His research continues to inspire students and engineers aiming to build adaptive, cooperative, and computationally tractable AI.