I am a Ph.D. student in the School of Intelligence Science and Technology at Peking University. Since March 2026, I have been advised by Prof. Yiwu Zhong; before that, I was advised by Prof. Yitao Liang. My research focuses on multi-modal reinforcement learning, embodied AI, and agentic models for Minecraft and other interactive environments.

Before joining Peking University, I received my B.E. from the Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University. I have also worked as a research intern with Prof. Zhiyuan Liu’s group at Tsinghua University and as a summer researcher at the University of Washington with Prof. Sheng Wang.

Research Interests

• Multi-modal reinforcement learning
• Embodied AI and open-ended agents
• Vision-language-action models
• Minecraft agents and interactive environments

News

• 2026: CrossAgent was accepted to CVPR 2026 as a Highlight.
• 2026: OpenHA and DeepHA were accepted to ICML 2026.
• 2025: MCU was accepted to ICML 2025 as a Spotlight.
• 2025: JARVIS-VLA was accepted to ACL Findings 2025.

Selected Publications

• CrossAgent: Bridging Cross-level Actions into One Agentic Model via Reinforcement Learning. CVPR 2026 Highlight.
• OpenHA: A Series of Open-Source Hierarchical Agentic Models in Minecraft. ICML 2026.
• DeepHA: Scaling Action Chains Elicits Deep Hierarchical Agents. ICML 2026.
• MCU: An Evaluation Framework for Open-Ended Game Agents. ICML 2025 Spotlight.
• JARVIS-VLA: Post-Training Large-Scale Vision Language Models to Play Visual Games with Keyboards and Mouse. ACL Findings 2025.

Contact

Primary Email: hkc20@stu.pku.edu.cn