About me
I am Yuchen Zhu (朱宇辰), a Ph.D. student at University of Science and Technology of China (USTC), jointly trained with Eastern Institute of Technology, Ningbo (EIT), advised by Prof. Wentao Zhu. Previously, I received my Master’s degree from ShanghaiTech University (2023–2026), advised by Prof. Sibei Yang, and my Bachelor’s degree from Nanjing Institute of Technology in 2022.
My research interests include Vision-Language-Action (VLA) models and Latent World Models. I am currently focusing on building embodied agents that can perceive, reason, and act in the physical world through scalable VLA systems and latent world modeling.
🔥 News
- 2026.10: 🎉 We introduce Juno, a unified predictive-latent framework for VLA pretraining, policy learning, and test-time adaptation. [Paper] [Code]
- 2026.07: 🎉🎉 I will join USTC as a Ph.D. student, jointly trained with EIT, Ningbo, advised by Prof. Wentao Zhu!
- 2025.09.19: 🎉🎉 Our paper “Intervene-All-Paths: Unified Mitigation of LVLM Hallucinations across Alignment Formats” has been accepted to NeurIPS 2025!
- 2025.07: 🎉🎉 Our paper “Sim-DETR: Unlock DETR for Temporal Sentence Grounding” has been accepted to ICCV 2025!
📝 Publications
* denotes equal contribution
Juno: Taming Predictive Latents for Vision-Language-Action Models
Yuchen Zhu, Chenyi Xu, Yulin Zhang, Gang Xu, Wentao Zhu
One action-conditioned JEPA for control-aligned pretraining, decoupled predictive reasoning, and dynamics-first test-time adaptation.
Rethinking Query-based Transformer for Continual Image Segmentation
Yuchen Zhu*, Cheng Shi*, Dingyou Wang, Jiajin Tang, Zhengxuan Wei, Yu Wu, Guanbin Li, Sibei Yang
Intervene-All-Paths: Unified Mitigation of LVLM Hallucinations across Alignment Formats
Jiaye Qian*, Ge Zheng*, Yuchen Zhu, Sibei Yang
Sim-DETR: Unlock DETR for Temporal Sentence Grounding
Jiajin Tang*, Zhengxuan Wei*, Yuchen Zhu, Cheng Shi, Guanbin Li, Liang Lin, Sibei Yang
❤️ Hobbies
I’m a big ⚽ football fan and love playing. I also play guitar for fun—back in college, I was in a band with friends!