東京大学松尾・岩澤研究室特任助教．ロボット学習とサービスロボティクスを専門としています．
2025年からは，汎用ロボットのオープンプラットフォーム開発に取り組む一般社団法人AIロボット協会（AIRoA）のCTOを務めています．
興味・関心
- ロボット学習
- サービスロボティクス
- 生成モデル
- 強化学習
学歴
博士（工学）, 2023
東京大学
修士（工学）, 2020
東京大学
学士（工学）, 2018
東京大学
最新
- Continuous Reasoning for Vision-Language-Action
- FlexLAM: Resolving the Bottleneck Trade-off in Latent Action Learning
- RAGDP: Retrieve-Augmented Generative Diffusion Policy
- See Less, Specify More: Visual Evidence Budgets for Generalizable VLAs
- Toward Hierarchical Vision-Language-Action Models with Continuous Skill Abstraction
- A comprehensive survey on physical risk control in the era of foundation model-enabled robotics
- Leave no observation behind: Real-time correction for vla action chunks
- Retrieve-Augmented Generation for Speeding up Diffusion Policy without Additional Training
- SPARK: Graph-Based Online Semantic Integration System for Robot Task Planning
- Data-driven Development Process in Service Robot Systems-Discussion Based on a Case Study from the World Robot Summit 2020 Partner Robot Challenge
- Gendom: Generalizable one-shot deformable object manipulation with parameter-aware policy
- Real World Applications of Large Language Models.
- Real-world robot applications of foundation models: A review
- Self-recovery prompting: Promptable general purpose service robot system with foundation models and self-recovery
- Special issue on real-world robot applications of the foundation models
- Collective intelligence for 2d push manipulations with mobile robots
- GenORM: Generalizable One-shot Rope Manipulation with Parameter-Aware Policy
- M3IL: Multi-Modal Meta-Imitation Learning
- TRAIL Team Description Paper for RoboCup@ Home 2024
- Weblab xArm dataset
- 大規模言語モデルを用いて意味的情報を考慮した物体把持位置の検出
- World robot challenge 2020--partner robot: a data-driven approach for room tidying with mobile manipulator
- Co-adaptation of algorithmic and implementational innovations in inference-based deep reinforcement learning
- Deployment-efficient reinforcement learning via model-based offline optimization
- Policy information capacity: Information-theoretic measure for task complexity in deep reinforcement learning
- Tool as embodiment for recursive manipulation
- Modeling task uncertainty for safe meta-imitation learning
- Neuron as an agent