Spatial Intelligence
Spatial intelligence systems for open urban spaces, connecting multimodal reasoning, world modeling, physical AI, and real-world applications.
I am an Assistant Professor at Zhongguancun Academy. My research lies at the intersection of spatial intelligence, agentic intelligence, and urban science.
My work moves beyond passive urban analytics toward spatial agents that can perceive multimodal environments, reason over geospatial context, simulate human behavior, and support decisions in open, dynamic cities.
I have published 40+ papers at venues including KDD, ICLR, ICCV, and ACL, with 5000+ citations and recognition from UbiComp, INFORMS, Stanford Top 2%, Tsinghua Shuimu Scholar, and ACM SIGSPATIAL China.
I received my B.S. and Ph.D. degrees in electronic engineering from Tsinghua University in 2016 and 2021, respectively, under the supervision of Prof. Yong Li. Before joining ZGC Academy, I was a Postdoctoral Fellow at Tsinghua University and a researcher at Meituan.
Learn moreSpatial intelligence systems for open urban spaces, connecting multimodal reasoning, world modeling, physical AI, and real-world applications.
Socially embodied, multimodal, and multi-agent systems for urban simulation, collaborative reasoning, and real-world decision support.
Urban analytics and human-behavior modeling for understanding activity patterns, mobility dynamics, and city-scale social systems.
UrbanWell was accepted to the D&B Track and OpenCity was accepted to the AI4S Track.
DynaOD was accepted by IJCAI 2026. Congratulations to Jie Zhao and Xianqi Dai.
Our survey on AI agent behavioral science was published in Humanities and Social Sciences Communications.
CityLens was accepted by ICLR 2026. Congratulations to Tianhui Liu.
We released a new survey about embodied world models with a regularly updated paper list.
Our work on mitigating geospatial knowledge hallucination was accepted by EMNLP 2025 Findings.
Our survey on world models was published in ACM Computing Surveys, with a regularly updated paper list.
Basic Spatial Abilities and PIGEON were accepted by ACL 2025.
Selected as one of the 2025 Spatial Data Intelligence Rising Stars.
Released a preprint on urban simulation with massive LLM-based agents.
Selected as one of the Stanford/Elsevier Top 2% Scientists 2024.
High Efficiency Delivery Network and UniST were accepted by KDD 2024.
Released a preprint on Urban Generative Intelligence in the era of large language models.
Meituan's real-time intelligent dispatching algorithms were selected as an INFORMS 2023 Edelman Award finalist.