Research
A connected program of scholarly inquiry
MSU Wood AI develops practical artificial intelligence and machine-learning methods for wood science, bioproducts, materials research, and industry applications.
Research framework
Our research begins with scientific and industrial problems that depend on time-intensive manual analysis. We work with industry partners to translate those challenges into rigorous machine-learning and computational research questions.
Across our projects, we connect artificial intelligence and computer vision with domain knowledge in bioproducts and engineered wood, including image classification, moisture and thermal transport, and renewable and alternative energy systems.
Applied AI and Industrial Automation
MSU Wood AI works with industry partners to identify research problems within workflows that rely on manual analysis or classification. The lab studies how artificial intelligence and machine learning can support more consistent, scalable ways to interpret data and images.
This work connects practical needs with model development and evaluation. The goal is to create research tools that complement domain expertise and can be assessed for the conditions in which they will be used.
- Industry collaboration
- Workflow automation
- Automated inspection
Questions
- Which labor-intensive tasks are suitable for reliable AI-assisted analysis?
- How can models be evaluated for the data and operating conditions of an industrial workflow?
Methods
- Machine learning
- Model development and validation
- Applied data analysis
Computer Vision and Drone Imagery
The lab develops and evaluates computer-vision methods for image-based classification problems. Drone imagery is one application, alongside other visual datasets that can support research, inspection, and materials assessment.
This area considers how data quality, class definitions, and evaluation affect whether an image classifier provides useful and repeatable results.
- Drone-image classification
- Computer vision
- Visual data analysis
Questions
- How can drone and other image datasets be classified accurately and consistently?
- How do image conditions and dataset composition affect model performance?
Methods
- Deep learning
- Image preprocessing and classification
- Performance evaluation
Bio-based Materials and Engineered Wood
The lab studies bio-based composites, engineered wood products, and the material properties that influence their performance. This work brings materials research together with computation, measurement, and data-driven analysis.
Artificial intelligence and machine learning are used where they can help characterize materials, interpret images or measurements, and support quality assessment.
- Bio-based composites
- Engineered wood products
- Materials assessment
Questions
- How can data-driven methods improve the assessment of wood and bio-based materials?
- Which material features are most useful for classification or property prediction?
Methods
- Materials characterization
- Machine learning
- Image-based assessment
Transport Modeling and Renewable Systems
The lab investigates moisture and thermal transport in materials and systems, including the ways heat and moisture move and interact. Modeling these processes supports a better understanding of performance under changing conditions.
Related work examines renewable and alternative energy systems, connecting computational analysis with questions in sustainable materials and energy use.
- Moisture transport
- Thermal transport
- Renewable and alternative energy
Questions
- How do coupled moisture and thermal processes influence material behavior?
- How can computational models support the analysis of renewable and alternative energy systems?
Methods
- Computational modeling
- Heat and mass transfer analysis
- Data analysis
Industry collaboration
From labor-intensive workflows to practical research tools
We collaborate with companies to study where artificial intelligence and machine learning can support repeatable analysis, classification, and inspection. Drone-image classification is one example of the image-based problems addressed through this work.