2026.06
Started research internships at Columbia University with Prof. Shilong Liu and at InSpatio, focusing on world models and embodied AI.
Yuhang Chen
Undergraduate Researcher
Huazhong University of Science and Technology
World models, multimodal intelligence, and embodied AI.
「己の珠に非ざることを惧れるが故に、敢えて刻苦して磨こうともせず、又、己の珠なるべきを半ば信ずるが故に、碌々として瓦に伍することも出来なかった。」Fearing that I might not be a jewel, I dared not labor to polish myself; yet half-believing that I was one, I could not bring myself to idle among common tiles.— Nakajima Atsushi, Sangetsuki (The Moon over the Mountain)
Fall 2027 · PhD Opportunities
I am interested in PhD opportunities for Fall 2027 in the United States, Singapore, and Hong Kong.
Get in touchAbout
Research Profile
I am a final-year undergraduate in Computer Science and Technology at Huazhong University of Science and Technology, expecting to graduate in June 2027. I am currently a research intern at Columbia University, working with Prof. Shilong Liu, and at InSpatio. My research focuses on interactive world models, spatial reasoning, and unified multimodal models for embodied AI. Previously, I worked with Prof. Tianyi Zhou at the University of Maryland and Prof. Yao Wan at ONE Lab, HUST.
- Interactive world models
- Spatial reasoning
- Unified multimodal models
- Embodied AI
News
Recent Updates
2026.04
Paper2Web was accepted to ACL 2026.
2026.02
GapEval was accepted to CVPR 2026 Findings.
2026.01
GapEval was released as a preprint.
2025.10
Paper2Web was released as a preprint.
2025.03
Judge Anything was accepted to KDD 2025 as an oral presentation.
2024.11
Judge Anything was released as a preprint.
2023.09
Began undergraduate study at HUST.
Research
Publications & Manuscripts
DIVE
DIVE: Benchmarking Full-Duplex Interaction with Visual Engagement in Continuous Streams
A benchmark for real-time full-duplex multimodal interaction, with 525 audio-visual samples across six tasks covering proactive intervention, continuous engagement, and interaction coordination.
In submissionarXiv
MOSAIC: Multi-Policy Sampling and Collaborative Test-Time Adaptation for Vision-Language-Action Models
Collaborative test-time adaptation for VLA policies, improving 103 of 104 cross-policy evaluations across four robotic manipulation benchmarks.
In submission
Background
Experience & Education
2026.06 - Present
Columbia University
Research intern with Prof. Shilong Liu. Interactive world models for perception, spatial reasoning, planning, and embodied interaction.
2026.06 - Present
InSpatio
Research intern in world models and embodied intelligence. Contributed to InSpatio-World 1.5 through training-data preprocessing and model training; research on world action models.
2025.09 - 2026.01
University of Maryland
Research intern with Prof. Tianyi Zhou. Token-efficient reasoning, user-centric AI, and the gap between understanding and generation in unified multimodal models.
2024.10 - 2026.02
ONE Lab, HUST
Research intern with Prof. Yao Wan. Academic webpage generation, multimodal consistency evaluation, and multimodal agents.
2023.09 - 2027.06
Huazhong University of Science and Technology
B.E. in Computer Science and Technology (expected June 2027). GPA: 4.42/5; rank: 18/313 (top 5.7%).
Honors
Scholarships and competitions
Outstanding Academic Performance Scholarship (2023, 2024, 2025); Technology Innovation Scholarship (2024, 2025); First Prize, National Finals of the Global Campus Artificial Intelligence Algorithm Elite Competition (2025); Third Prize, Computer System Development Capability Competition (2025); Second Prize, RuiKang Robot Developer Competition (2025).
Japan
A Personal Travel Record
Here are photos I took while traveling in Japan. I love traveling in Japan - the people, the scenery, and the food. I hope to study there someday and eventually settle down in this wonderful and fascinating country.
01 / 67