Daisuke Uragami

Tokyo University of Technology, Tokyo Denki University, Nihon University

Papers

Total Citations

H-Index

About

Daisuke Uragami’s research sits at the intersection of cognitive science and robotics, exploring how human-like decision-making can revolutionize machine learning. His primary focus is on cognitively inspired reinforcement learning, where he investigates how cognitive biases—often seen as human flaws—can actually enhance robotic motion learning and control. In his most cited work, "Cognitively Inspired Reinforcement Learning Architecture and Its Application to Giant-Swing Motion Control" (2013, 8 citations), Uragami demonstrates how embedding human cognitive shortcuts into algorithms enables robots to master complex physical tasks, like a giant-swing motion, with remarkable efficiency. His 2016 paper "Cognitive Satisficing" (6 citations) tackles a critical challenge in reinforcement learning: the explosion of state-action spaces. By introducing a satisficing approach—choosing "good enough" solutions rather than optimal ones—Uragami shows how agents can learn effectively under severe physical and computational constraints. Further, his work on symmetric cognitive biases (2011, 3 citations) and coarse-grained state spaces (2016, 2 citations) reveals how illogical human tendencies can be systematically harnessed to accelerate robotic learning. While his citation counts are modest, Uragami’s contributions are pioneering, offering a fresh, cognitively grounded pathway for developing more adaptive and resource-efficient autonomous systems.