Tatsuya Matsushima is a Project Assistant Professor at The University of Tokyo, specializing in robot learning and service robotics.
Since 2025, he has been the CTO of AIRoA, a non-profit organization developing open platforms for generalist robots.
Interests
- Robot Learning
- Service Robotics
- Generative Models
- Reinforcement Learning
Education
PhD in Engineering, 2023
The University of Tokyo
MS in Engineering, 2020
The University of Tokyo
BS in Engineering, 2018
The University of Tokyo
Experience
Project Assistant Professor
Research on robot learning and service robotics.
Selected Publications
(2021).
Deployment-efficient reinforcement learning via model-based offline optimization.
International Conference on Learning Representations.
Recent Publications
(2026).
Continuous Reasoning for Vision-Language-Action.
arXiv preprint arXiv:2606.00229.
(2026).
FlexLAM: Resolving the Bottleneck Trade-off in Latent Action Learning.
arXiv preprint arXiv:2606.19408.
(2026).
See Less, Specify More: Visual Evidence Budgets for Generalizable VLAs.
arXiv preprint arXiv:2606.02735.
(2026).
Toward Hierarchical Vision-Language-Action Models with Continuous Skill Abstraction.
ICRA 2026 Workshop: From Data to Decisions: VLA Pipelines for Real Robots.
(2026).
YUBI: Yielding Universal Bidigital Interface for Bimanual Dexterous Manipulation at Scale.
arXiv preprint arXiv:2606.10244.
(2026).
推論遅延下におけるリアルタイム応答性のための VLA アクションチャンクの補正.
人工知能学会全国大会論文集 第 40 回 (2026).
(2026).
私のブックマーク: ロボット領域におけるフィジカル AI.
人工知能.