Ruifeng Luo罗睿锋

Chief Technology Officer, Arctron · Arcplus Group
Ph.D. in Civil Engineering, Tongji University — working on AI for Engineering

Date 2026.10
Rev B — UPDATED
Location Shanghai, CN
Sheet 1 OF 1
1 2 3 A B 7200 7200 14400 OFFICE 21.5 m² MEETING 21.5 m² A-WALL A-GLAZ A-DOOR N
Fig. 0 — What my models learn to read: primitives, symbols, layers.

I am the Chief Technology Officer of Arctron, the digital-technology subsidiary of Arcplus Group, where I lead a cross-disciplinary team in Shanghai building AI for the built environment. My research asks one question: can machines read engineering drawings the way engineers do?

Toward that goal, I work on vector multimodal models for CAD drawings, large-scale open datasets (ArchCAD-400K, NeurIPS 2025), and domain benchmarks for large language models (AECBench, Advanced Engineering Informatics). Earlier, my Ph.D. work at Tongji University applied reinforcement learning and Monte-Carlo tree search to structural design. I also serve as an adjunct graduate supervisor at Tongji University.

Ruifeng Luo

Research · by layer

L-01 · DRAWING-INTEL

Vector multimodal models for engineering drawings

Teaching models to parse CAD natively — panoptic symbol spotting, vector-line representations, and the data engines that make large-scale annotation affordable.

L-02 · AEC-LLM-EVAL

Evaluating LLMs in the AEC domain

Hierarchical benchmarks that measure what engineering practice actually demands: knowledge memorization, understanding, reasoning, calculation, and application.

L-03 · STRUCT-GEN

Intelligent structural design

Reinforcement learning and Monte-Carlo tree search for generative structural design, from truss layout to topology optimization.

News

Selected publications

Automation in Construction 2027Corresponding author

Vector-level panoptic symbol spotting for CAD drawing understanding

Zhengjie Liu, Jianing Zhang, Zhaoqi Huang, Yujie Lu, Ruifeng Luo, and Xianzhong Zhao.
Automation in Construction, 193, 107286 (2027).

SegCAD represents CAD primitives as point clouds for vector-level panoptic symbol spotting. The ArchCAD benchmark contains 40,000 verified samples across 30 categories; SegCAD achieves 82.7% PQ, with a 3D structural reconstruction case study demonstrating downstream use.

Paper DOI
BibTeX 
@article{LIU2027107286,
 author = {Zhengjie Liu and Jianing Zhang and Zhaoqi Huang and Yujie Lu and Ruifeng Luo and Xianzhong Zhao},
 title = {Vector-level panoptic symbol spotting for CAD drawing understanding},
 journal = {Automation in Construction},
 volume = {193},
 pages = {107286},
 year = {2027},
 issn = {0926-5805},
 doi = {10.1016/j.autcon.2026.107286},
 url = {https://www.sciencedirect.com/science/article/pii/S0926580526005273}
}
NeurIPS 2025First author

ArchCAD-400K: A Large-Scale CAD Drawings Dataset and New Baseline for Panoptic Symbol Spotting

413K annotations across 27 categories from 5,538 standardized drawings — 26× larger than the largest prior CAD dataset — with a structure-aware annotation engine and the DPSS baseline.

arXiv Project Code Data
BibTeX 
@inproceedings{NEURIPS2025_b96ce7d3,
 author = {Luo, Ruifeng and Liu, Zhengjie and Cheng, Tianxiao and Wang, Jie and Wang, Tongjie and Cheng, Fei and Chai, Fu and Li, Yanpeng and Wei, Xingguang and Wang, Haomin and Ye, Shenglong and Wang, Wenhai and Zhang, Zhang and Qiao, Yu and Zhang, Hongjie and Zhao, Xianzhong},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {D. Belgrave and C. Zhang and H. Lin and R. Pascanu and P. Koniusz and M. Ghassemi and N. Chen},
 pages = {127715--127739},
 publisher = {Curran Associates, Inc.},
 title = {ArchCAD-400K: A Large-Scale CAD drawings Dataset and New Baseline for Panoptic Symbol Spotting},
 url = {https://proceedings.neurips.cc/paper_files/paper/2025/file/b96ce7d38339874a8704e8895f743284-Paper-Conference.pdf},
 volume = {38},
 year = {2025}
}
NeurIPS 2025Co-author

Point or Line? Using Line-based Representation for Panoptic Symbol Spotting in CAD Drawings

VecFormer: a line-based representation of vector primitives for panoptic symbol spotting.

PDF
BibTeX 
@inproceedings{NEURIPS2025_4791edcb,
 author = {Wei, Xingguang and Wang, Haomin and Ye, Shenglong and Luo, Ruifeng and Zhang, Zhang and Gu, Lixin and Dai, Jifeng and Qiao, Yu and Wang, Wenhai and Zhang, Hongjie},
 booktitle = {Advances in Neural Information Processing Systems},
 editor = {D. Belgrave and C. Zhang and H. Lin and R. Pascanu and P. Koniusz and M. Ghassemi and N. Chen},
 pages = {50036--50061},
 publisher = {Curran Associates, Inc.},
 title = {Point or Line? Using Line-based Representation for Panoptic Symbol Spotting in CAD Drawings},
 url = {https://proceedings.neurips.cc/paper_files/paper/2025/file/4791edcba96fbd82a8962b0f790b52c9-Paper-Conference.pdf},
 volume = {38},
 year = {2025}
}
Adv. Eng. Informatics 2026Corresponding author

AECBench: A Hierarchical Benchmark for Knowledge Evaluation of Large Language Models in the AEC Field

4,800 questions, 23 tasks, five cognitive levels — the first hierarchical benchmark of LLM knowledge for architecture, engineering and construction.

arXiv Code
BibTeX 
@article{Liang_2026,
 author = {Liang, Chen and Huang, Zhaoqi and Wang, Haofen and Chai, Fu and Yu, Chunying and Wei, Huanhuan and Liu, Zhengjie and Li, Yanpeng and Wang, Hongjun and Luo, Ruifeng and Zhao, Xianzhong},
 title = {AECBench: A hierarchical benchmark for knowledge evaluation of large language models in the AEC field},
 journal = {Advanced Engineering Informatics},
 volume = {71},
 pages = {104314},
 year = {2026},
 month = apr,
 issn = {1474-0346},
 doi = {10.1016/j.aei.2026.104314},
 url = {http://dx.doi.org/10.1016/j.aei.2026.104314},
 publisher = {Elsevier BV}
}
Buildings 2022First author

AlphaTruss: Monte Carlo Tree Search for Optimal Truss Layout Design

MCTS for truss layout under stress, displacement and buckling constraints.

Paper
BibTeX 
@article{Luo_2022,
 author = {Luo, Ruifeng and Wang, Yifan and Xiao, Weifang and Zhao, Xianzhong},
 title = {AlphaTruss: Monte Carlo Tree Search for Optimal Truss Layout Design},
 journal = {Buildings},
 volume = {12},
 number = {5},
 pages = {641},
 year = {2022},
 month = may,
 issn = {2075-5309},
 doi = {10.3390/buildings12050641},
 url = {http://dx.doi.org/10.3390/buildings12050641},
 publisher = {MDPI AG}
}

Full list on Google Scholar.

Academic Service