Takuto Sakuma
Papers
4
Total Citations
22
H-Index
3
About
Takuto Sakuma is a researcher at the intersection of machine learning, robotics, and computational biology. His work is defined by a dual focus: developing efficient algorithms for pattern recognition in complex sequential data, and advancing autonomous mobile robotics through reinforcement learning. In a notable 2019 study (10 citations), Sakuma introduced an efficient learning algorithm for sparse subsequence pattern-based classification, applying it to the novel domain of comparative animal trajectory data analysis—a method that helps biologists extract meaningful behavioral insights from high-resolution movement logs. On the robotics side, Sakuma has been a driving force in sim-to-real transfer learning, tackling the critical challenge of deploying policies trained in simulation onto physical hardware. His 2021 paper on this topic (6 citations) and his 2020 work on policy transfer for omnidirectional robots (4 citations) demonstrate practical frameworks for closing the reality gap, enabling warehouse robots to navigate and avoid obstacles using LiDAR and deep reinforcement learning. With additional contributions to path planning using Deep Deterministic Policy Gradient, Sakuma’s research offers a compelling blueprint for bridging data-driven algorithms and real-world autonomous systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3
- 4