Yihang Tao

Yihang Tao | 陶祎航

PhD Student

City University of Hong Kong

Research Interests

  • Autonomous Driving
  • Spatial Intelligence
  • Generative Model
  • AI Security

About Me

I received my B.S. degree from the School of Information Science and Engineering, Southeast University, Nanjing, China, in 2021, and my M.S. degree from the School of Computer Science, Shanghai Jiao Tong University, Shanghai, China, in 2024. Currently, I am pursuing my PhD degree in the Department of Computer Science at City University of Hong Kong (Advisor: Prof. Yuguang "Michael" Fang). My research broadly focuses on trustworthy Spatial Intelligence for Embodied AI, particularly in Autonomous Driving.

I am actively seeking collaborators of like-minded interests. If you are interested in my work, please feel free to contact me via email: yihang.tommy@my.cityu.edu.hk.

🚀 Research Thrust

"Imagine Embodied AI that perceives the world not just as a stream of data, but as a coherent reality. I envision endowing agents with human-like Spatial Intelligence to construct a dynamic World Model (or Digital Twin), seamlessly bridging the gap between digital simulation and physical reality."

To achieve this vision, my research aims to construct a comprehensive World Model for Spatial Intelligence, serving as the cognitive core for Embodied AI (e.g., Autonomous Vehicles). This system not only mirrors the current physical world but also reasons about its dynamics. I approach this goal through two complementary thrusts:

Collaboration for Extended 3D Understanding

This thrust focuses on extending the spatial breadth of the World Model. By establishing Collaborative Perception among agents, it overcomes individual sensing limitations (e.g., occlusions, limited range). This collective intelligence builds a more complete and robust World Model, ensuring agents maintain accurate 3D understanding even in complex, adversarial, or bandwidth-constrained environments.

Generation for Predictive 3D Understanding

This thrust focuses on extending the temporal depth of the World Model. By leveraging Generative Models and 3D Gaussian Splatting, it empowers agents to hallucinate plausible future scenarios and recover 3D geometry from sparse observations. This predictive capability enables the World Model to simulate potential outcomes, facilitating foresightful planning and decision-making for Embodied AI.

🔥 News

View All

📄 Publications

(* means equal contribution)

🏆 Honors & Awards

View All

💼 Academic Service

Conference Reviewer

Program Committee Member

ECCV (2026), CVPR (2026), ICASSP (2026), ACM MM (2025), ICML (2025-2026), ICLR (2025), IJCNN (2025), IEEE ICC (2025), IEEE GLOBECOM (2023-2025), IEEE ICCC (2024)

Journal Reviewer

Peer Review

IEEE TMC, IEEE TITS, IEEE TCE, Pattern Recognition, Neural Networks, EAAI, IEEE JBHI, IEEE LNET

Session Chair

IEEE GLOBECOM 2023

MWN Track, Semantic Communications Session

Teaching Assistant

City University of Hong Kong

• CS5279 Topics in AI Security (2026 Spring)
• CS5222 Computer Networks & Internets (2025 Fall)
• CS1302 Introduction to Computer course (2024 Fall, 2025 Spring)

🎓 Education

CityU

PhD @ City University of Hong Kong

Sep. 2024 - Now

Department of Computer Science

Supervisor: Prof. Yuguang "Michael" Fang

SJTU

MEng @ Shanghai Jiao Tong University

Sep. 2021 - Apr. 2024

School of Computer Science

SEU

BEng @ Southeast University

Sep. 2017 - Jul. 2021

School of Information Engineering

📍 Contact

Office Address

83 Tat Chee Avenue Kowloon, Hong Kong

Visitor Statistics