Kenji Sugimoto
Nara Institute of Science and Technology, Graphic Era University
Papers
19
Total Citations
293
H-Index
9
About
Kenji Sugimoto’s research lies at the intersection of machine learning, robotics, and human-robot interaction, with a focus on developing algorithms that enable robots to learn robustly from noisy data and adapt to dynamic environments. His most influential work, “Robust Stochastic Gradient Descent With Student-t Distribution Based First-Order Momentum” (67 citations), introduces a novel optimization method that improves the stability of deep neural network training by mitigating the impact of noisy supervision signals—a critical contribution for real-world applications. In robotics, Sugimoto has pioneered reinforcement learning approaches for pneumatic artificial muscle-driven robots (38 citations) and kernel-based policy programming for high-dimensional state spaces (29 citations), enabling more efficient motor skill acquisition. His work on active tactile object recognition (32 citations) and environment-adaptive interaction primitives (18 citations) advances how robots perceive and collaborate with humans through touch and visual context. Notably, his research on intelligent mobility aids (16 citations) and care motion controllers (10 citations) demonstrates a commitment to assistive robotics, aiming to improve quality of life for individuals with physical or cognitive impairments. With over 260 total citations, Sugimoto’s contributions are shaping the next generation of adaptive, human-centered robotic systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10