Bo Xiao
Michigan Technological University, MI·Engineering
Construction sites generate huge amounts of visual and digital information, yet construction is the least digitized of the major US industries, with persistent problems of low productivity, poor coordination and a workforce short on digital skills. Two projects address different ends of this gap: one builds shared open-source data and tools so that artificial intelligence can be trained for construction tasks, and the other builds AI-driven teaching tools so that future construction engineers learn the digital technologies the industry needs.
- Fields
Engineering
Computer Science
Education Research
- Topics
construction engineering
artificial intelligence
computer vision
- Grants
- 2 known grants, through Aug 2027
What they work on
- Analyze the current technological landscape of construction AI data and survey the community to identify construction-specific capability gaps and growth directions for the Advanced Construction Image Dataset suite.
- Evaluate a hierarchical consortium governance model for an open-source construction AI ecosystem, aiming to sustain data quality, ethical compliance and contributor engagement across academia, industry and government.
- Develop a security-by-design framework through which datasets, models, applications and documentation can be validated and shared among distributed contributors under defined data-quality and privacy rules.
- Create undergraduate curriculum materials for building information modeling (BIM), digital representations of a facility's physical and functional characteristics, organized using knowledge space theory from cognitive science and vetted by expert discussion.
- Develop a generative design method combining parametric modeling and computational techniques to automatically produce BIM worked examples at multiple difficulty levels for instructors.
- Build and experimentally validate a ChatGPT-based web application that delivers personalized, adaptive and interactive BIM training to undergraduate construction students.
More about the research
The first project grows the Advanced Construction Image Dataset suite, a widely used collection of construction-specific AI datasets covering object detection (locating things in images), image captioning (describing images in words) and instance segmentation (outlining each individual object), together with annotation guidelines and benchmark analyses, currently serving over 700 active users. The aim is to turn it into a decentralized, self-sustaining open-source ecosystem offering datasets, models, algorithms, documentation and learning materials, and linking academia, industry, government and professional societies. Work in this phase includes a systematic analysis of the current construction data landscape, an assessment of construction-specific capability gaps, and a broad community survey on future directions. The project also evaluates whether a hierarchical consortium governance model can sustain quality, ethical compliance and contributor engagement, and develops a security-by-design framework so datasets, models and applications can be validated and shared among distributed contributors under clear data-quality and privacy rules. Community building proceeds through workshops, competitions and international collaborations.
Why it matters
Construction suffers from low productivity and a shortage of digitally skilled workers. A sustained open-source ecosystem of construction image datasets, models and governance could let companies, agencies and researchers build and share AI tools for site monitoring and safety rather than each starting from scratch. The BIM training platform could make digital-modeling instruction available where qualified instructors are scarce, helping address engineering labor shortages and reaching underrepresented students in Michigan and the Midwest.
Known grants
From NIH and NSF research grants. Labs often have other funding we can't see.
Closest labs
Jee Woong Park
University of Nevada-Las Vegas, NV·Engineering
Develop computer vision AI algorithms that detect workplace hazards in photographs that trainees take on their own construction sites.
Aditya Johri
George Mason University, VA·Education Research
Create four role-play case studies on AI-enabled security technologies, such as video surveillance, smart home privacy, and algorithmic hiring, and embed them across undergraduate and graduate cybersecurity courses.
Tianqi Chen
Carnegie Mellon University, PA·Computer Science·Early-career
Build a unified program representation that captures both high-level computation and data formats and low-level hardware optimizations of AI models, as the foundation for a new machine learning compiler.
The school
Michigan Technological University: 9 labs on PI Pages, $107M a year on research, 86 research doctorates a year.
Reach out
A start for your email. Fill in the brackets.