Riku Narita

Papers

Total Citations

H-Index

About

Riku Narita is a robotics researcher advancing autonomous decision-making in complex environments through reinforcement learning. Their primary research focuses on heterogeneous multi-agent reinforcement learning (HMARL) for single-robot systems, addressing the critical challenge of efficient exploration and adaptation to dynamic environments. Narita’s most cited work, "Effective Action Learning Method Using Information Entropy for a Single Robot Under Multi-Agent Control" (2024, 2 citations), introduces a novel approach where multiple learning agents operate within a single robot, using information entropy to guide action selection and improve learning efficiency in changing conditions. Their earlier paper, "Efficient exploration by switching agents according to degree of convergence of learning on Heterogeneous Multi-Agent Reinforcement Learning in Single Robot" (2021, 2 citations), tackled the problem of random exploration in conventional reinforcement learning by developing a switching mechanism that selects agents based on learning convergence, enabling more targeted and efficient exploration. While still early in their career, Narita’s work represents a promising direction in integrating multi-agent learning frameworks within single robotic platforms, potentially reducing computational overhead while improving adaptability. Their research has implications for autonomous robots operating in unpredictable real-world environments, from search-and-rescue to industrial automation.