Shin‐ichi Maeda
Preferred Networks (Japan), Vector Institute, University of Toronto
Papers
8
Total Citations
89
H-Index
3
About
Shin‐ichi Maeda is a robotics researcher advancing the frontiers of reinforcement learning, robotic manipulation, and gripper design. His work bridges the gap between human intuition and autonomous systems, notably through DQN-TAMER (48 citations), a human-in-the-loop RL framework that tackles exploration challenges by integrating intractable human feedback—a critical step for real-world robotics applications. Maeda also addresses practical industrial needs with uncertainty-aware self-supervised grasping of granular foods (26 citations), enabling robots to adaptively pick target masses of diverse food items with minimal training. His hardware innovations include the F1 Hand, a fixed-finger gripper designed for delicate teleoperation and autonomous grasping, and the FAAF Hand, a four-axis adaptive fingers hand optimized for object insertion tasks. Through the MANGA framework, he decouples policy learning from system identification, allowing neural policies to generalize across varying dynamics and noise conditions. With a portfolio spanning from foundational RL algorithms to deployable grippers, Maeda’s work directly impacts automation in food packing, teleoperation, and fine-grained manipulation, demonstrating a commitment to robust, adaptive robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2Uncertainty-aware Self-supervised Target-mass Grasping of Granular Foods26 citations · 2021
- 3
- 4MANGA: Method Agnostic Neural-policy Generalization and Adaptation3 citations · 2019
- 5
- 6Four-Axis Adaptive Fingers Hand for Object Insertion: FAAF Hand2 citations · 2024
- 7MANGA: Method Agnostic Neural-policy Generalization and Adaptation2 citations · 2020
- 8Uncertainty-Aware Self-Supervised Target-Mass Grasping of Granular Foods2 citations · 2021