Here is Di Wu. I am a second-year masterโs student in the School of Computer Science at Tongji University and a member of the MAIL-TAII Lab, advised by Prof. Bo Jin and Wenhao Li
My Research Motivation: I believe Reinforcement Learning with prior knowledge (Foundation models) can make a more divergent and capable Agent for human life
Specifically, it includes:
-
Agentic RL: Developing foundation-model-based agents that can reason, plan, and act. I am particularly interested in reinforcement learning and self-evolution methods that enable agents to move beyond supervised learning and autonomously acquire new capabilities for real-world tasks.
-
Embodied Agents: Building agents that can perceive, reason, and act in the physical world, with a particular focus on self-evolving embodied models and agentic embodied systems.
-
Multi-Agent System: Exploring efficient methods to construct multi-agent systems for solving tasks, as well as the emergent social behaviors among multiple agents.
๐ฅ News
- ๐ 2026.4: One first-author paper has been accepted by ACL 2026 Main Track about Embodied Task Planning and RL Post-Training.
- ๐ 2025.9: Honored to receive the National Scholarship for Graduate Students at Tongji University (ranked 1st out of 63).
- ๐ 2025.4: One first-author survey paper has been accepted by IJCAI 2025 about Multi-agent Collaboration in Embodied AI.
- ๐ 2024.7: Start my research journey
๐ Educations
- 2020 - 2024: School of Software Engineering, Tongji University, Bachelor
- 2024 - 2027 (Expected): School of Computer Science, Tongji University, Master