Riku Arakawa

About

Riku Arakawa is a researcher at the forefront of human-robot interaction and reinforcement learning (RL), with a focus on making robot teaching more intuitive and efficient. His work bridges the gap between complex machine learning algorithms and practical, user-friendly robotic systems. Arakawa’s most cited paper, "DQN-TAMER: Human-in-the-Loop Reinforcement Learning with Intractable Feedback" (2018, 48 citations), addresses a core challenge in RL—exploration—by integrating human feedback to accelerate learning, a critical step for real-world robotics applications. He further advanced robot teaching with "A Multimodal Learning-from-Observation Towards All-at-once Robot Teaching using Task Cohesion" (2022), which enables robots to learn sequential tasks from combined language and demonstration inputs, moving beyond cumbersome step-by-step instructions. Demonstrating his versatility, Arakawa also pioneered the use of event cameras in RL for robotics, as shown in "Exploration of Reinforcement Learning for Event Camera using Car-like Robots" (2020), achieving faster control by leveraging the camera’s low latency. Through these contributions, Arakawa is shaping a future where robots can be taught more naturally and efficiently, reducing the barriers between human intent and robotic action.

Research Focus

Key Achievements

Top Papers

1. 1
   DQN-TAMER: Human-in-the-Loop Reinforcement Learning with Intractable Feedback

   48 citations · 2018
2. 2
   A Multimodal Learning-from-Observation Towards All-at-once Robot Teaching using Task Cohesion

   10 citations · 2022
3. 3
   Exploration of Reinforcement Learning for Event Camera using Car-like Robots

   4 citations · 2020

Key Collaborators

Contact & Links