Drag the glass drop to run the night into morning. Everything on the roof can be tapped.
Haihong Hao 郝海宏
I am a PhD student at the University of Science and Technology of China. My interests are models, agent systems and general intelligence in the digital world.
About

At USTC’s Future Media Computing Laboratory, I am supervised by Prof. Xiaojun Chang and my senior brother Mingfei Han. Before that, I was advised by Prof. Xiangnan He in the School of Data Science at USTC. My research interest lies in Vision-language navigation (VLN), 3D Large Language Models (3D-LLMs) and Multimodal-Large-Language-Models (MLLM). I am working hard to become an excellent researcher. If you have any good ideas or cooperation, please don’t hesitate to contact me.
I love coding, and I feel really happy when I get lost in it and solve the bugs I come across. I’m working hard to make sure my work does both: tells a good story and gets the job done. I also used to be into photography and I’m a League of Legends (LOL) mobile player, ranked at the highest level, King. I bought this domain, “abdd.top” in 2022 because I thought it was fun and cool.
News
One paper accepted by NeurIPS.
One paper accepted by TPAMI.
One paper accepted by ECCV.
One paper accepted by ICML.
One paper accepted by TPAMI.
One paper accepted by ICLR.
Champion of VLN-PE in the InternUtopia and Real World Challenge at IROS 2025.
One paper accepted by International Journal of Machine Learning and Cybernetics.
Publications
My published work spans vision-language navigation, multimodal understanding and geometric reasoning. More on Google Scholar.
GeoSense: Internalizing Geometric Necessity Perception for Multimodal Reasoning
Implicit Geometry Representations for Vision-and-Language Navigation from Web Videos
SELongVLM: Empowering Long Video Language Models with Self-Corrective Clip Selection
Progressive Online Video Understanding with Evidence-Aligned Timing and Transparent Decisions
Self-Consistency as a Free Lunch: Reducing Hallucinations in Vision-Language Models via Self-Reflection
CoNav: Collaborative Cross-Modal Reasoning for Embodied Navigation
Attribute encoding transformer on unattributed dynamic graphs for anomaly detection
International Journal of Machine Learning and Cybernetics
Hierarchical Space-Time Attention for Micro-Expression Recognition
Background
Education
Sep 2024 to present, Hefei
PhD in Computer Science
University of Science and Technology of China
Advisors: Xiaojun Chang, Mingfei Han
Sep 2022 to Jun 2024, Hefei
Master in Computer Science
University of Science and Technology of China
Advisor: Xiangnan He
Sep 2018 to Jun 2022, Zhengzhou
Bachelor in Computer Science
Zhengzhou University
Experience
Jun 2026 to present
HONOR
New Industry Incubation Department
Algorithm development for humanoid robot mobile manipulation.
Dec 2022 to Apr 2024
Chery HuiYin Motor Finance
Research
Tools and platforms
- Python, PyTorch, C/C++, MATLAB, embedded systems
- Data preparation, model training and evaluation, Sim2Real transfer
- Unitree G1 and Go2, AgileX LIMO, TurtleBot 4, Franka, UR, mobile manipulators
Academic Service
Reviewer: NeurIPS, CVPR, ICLR, TPAMI, ICRA, IROS, ICME.
Awards
-
2025
Champion, VLN-PE track
InternUtopia and Real World Challenge, IROS 2025. Prize: $11,500 (¥80,000).
Haihong Hao, Mingfei Han, Yi Xiang, Wei Ji, Xiaojun Chang
- 2025
China Scholarship Council (CSC) Scholarship
$2,000 (¥14,000) per month for 24 months.
- 2020
National Encouragement Scholarship, China
¥10,000.
Photos
Contact
If you have a good idea or would like to collaborate, please get in touch.



