Jianhua Han - Profile

🚀 About Me

I am currently a Research Scientist at 2030 Lab, Yinwang Intelligent Technology Co. Ltd. Previously, I worked as a Research Scientist at Huawei Noah's Ark Lab.

My research interests lie in Vision-Language Models (VLM), Vision-Language-Action (VLA) Models, World Models, and Physical AI (integrating Autonomous Driving & Robotics).

📍 Location: Shanghai, China
🎓 Education: M.S. (IEEE Class) & B.S. from Shanghai Jiao Tong University (SJTU).
📫 Contact: hanjianhua2012@gmail.com

📢 Hiring & Collaboration

We are actively looking for self-motivated Interns and Full-time Researchers/Engineers to join our team at 2030 Lab, Yinwang.

Focus Areas: VLM, Physical AI, VLA, World Models, End-to-End Autonomous Driving, and General-purpose Agents.
If you are interested in building the next generation of intelligent systems, please email me at hanjianhua2012@gmail.com.

🛠 Experience

Research Scientist @ 2030 Lab, Yinwang Intelligent Technology (2025 – Present)

Constructing research on Vision-Language Models (VLM), Vision-Language-Action (VLA) models and World Models for Physical AI.
Focusing on visual understanding, reasoning capabilities, and decision-making in complex open-world environments (AD & Robotics).

Research Scientist @ Huawei Noah's Ark Lab (2019 – 2025)

Foundation Models: Led the development of Large-scale Multi-modal Foundation Models (30B+ parameters).
- Deeply involved in model architecture design, pre-training strategies, and data optimization.
- The models achieved superior performance on benchmarks like MME, DocVQA, and MathVista.
Open-Vocabulary Perception: Developed the DetCLIP series, achieving No.1 results on COCO, Waymo 2D, and LVIS leaderboards.

📝 Selected Publications

1. Vision-Language-Action (VLA) & Embodied AI (Physical AI)

Focus: Integrating reasoning with action for Robotics and Autonomous Driving.

[New] Percept-WAM: Perception-Enhanced World-Awareness-Action Model for Robust End-to-End Autonomous Driving.
Focus: Integrating perception features into World Models to enhance robustness.
Reason2drive: Towards interpretable and chain-based reasoning for autonomous driving. (ECCV)
M Nie, R Peng, C Wang, X Cai, J Han, H Xu, L Zhang
Navcot: Boosting LLM-based vision-and-language navigation via learning disentangled reasoning. (TPAMI)
B Lin, Y Nie, Z Wei, J Chen, S Ma, J Han, H Xu, X Chang, X Liang
SeePhys: Does Seeing Help Thinking? – Benchmarking Vision-Based Physics Reasoning. (NeurIPS)
HiLM-D: Towards High-Resolution Understanding in Multimodal Large Language Models for Autonomous Driving. (IJCV)
X Ding, J Han, H Xu, W Zhang, X Li

2. Multimodal Foundation Models (Generation & Reasoning)

ILLUME: Illuminating Your LLMs to See, Draw, and Self-Enhance. (ICCV)
C Wang, G Lu, J Yang, R Huang, J Han, L Hou, W Zhang, H Xu
ILLUME+: Illuminating Unified MLLM with Dual Visual Tokenization and Diffusion Refinement. (In Submission)
SemHiTok: A Unified Image Tokenizer via Semantic-Guided Hierarchical Codebook. (In Submission)
Emova: Empowering language models to see, hear and speak with vivid emotions. (CVPR)
Gaining Wisdom from Setbacks: Aligning Large Language Models via Mistake Analysis. (ICLR)

3. Open-Vocabulary Perception (DetCLIP Series)

DetCLIP: Dictionary-enriched visual-concept paralleled pre-training for open-world detection. (NeurIPS)
L Yao*, J Han*, Y Wen, et al.
DetCLIPv2: Scalable Open-Vocabulary Object Detection Pre-training via Word-Region Alignment. (CVPR)
DetCLIPv3: Towards Versatile Generative Open-vocabulary Object Detection. (CVPR)
DetGPT: Detect what you need via reasoning. (EMNLP)