About

Somayeh Asadi is a Professor in the Department of Civil and Environmental Engineering at the University of Virginia. She previously served as an Associate Professor in the Department of Architectural Engineering at Pennsylvania State University. She holds M.Sc. and Ph.D. degrees in engineering science with an emphasis on construction engineering from Louisiana State University. Her research focuses on transforming smart infrastructure systems through research and education, with particular attention to smart buildings and cities, occupant behavior and health, net-zero carbon built environments, resilient and adaptive built environments, and the food–energy–water–society nexus. Her work also involves machine learning and artificial intelligence. Beyond her research, Dr. Asadi is dedicated to education and mentorship, fostering the development of both undergraduate and graduate students. She has served as the past chair of the ASCE Global Computing Committee and as an advisor for the National Electrical Contractor Association student chapter for eight years.

Research topics

• Computer Science
• Environmental economics
• Economics
• Engineering
• Political Science
• Distributed computing
• Artificial Intelligence
• Microeconomics
• Risk analysis (engineering)
• Pathology
• Knowledge management
• Business
• Mathematical optimization
• Operations management
• Electrical engineering
• Data science
• World Wide Web
• Internet privacy
• Medicine

Selected publications

• Human-centered energy modeling: integrating occupant behavior in simulation workflows

  Energy and Buildings · 2026-01-08

  articleSenior author
  PublisherDOI

• Investigating Instructors’ Experiences in a Neurodiversity-Focused AI Training Program

  2025-08-21

  article
  PublisherDOI

• Evaluation of Parameters that Impact Data Collection for Robot Progress Detection of a Masonry Wall

  Journal of Computing in Civil Engineering · 2025-11-25

  article

  The construction industry is going through a technological revolution where aspects of digitalization are combined with automation. A prime example is using data from a building information model (BIM) to promote robotic construction. However, numerous parameters impact the robot’s ability to collect data and provide user feedback for site progress detection. For instance, during progress detection, the robot must recognize site work features, such as a masonry wall. Therefore, this study investigates four parameters that would impact the data collection process for progress detection, including wall configuration, robot navigation path, image spacing, and distance from the wall. Photogrammetry is used to capture images with a teleoperated Husky A200 robot equipped with a stereo camera and the Global Positioning System (GPS). COLMAP was used to generate both the sparse and dense point cloud reconstruction and statistics related to the model, such as the reconstruction time, mean reprojection error, percentage of images used for reconstruction, and points generated. Once COLMAP processed all reconstructions, each model was reviewed individually. The results were filtered to determine viable methods for the data collection process and provide insight into which methods required further analysis to ascertain the quality of the model. However, the main goal of this study is to document and investigate the methods used in the data collection process while determining which methods should facilitate the optimal data collection process.

  PublisherDOI

• Optimizing thermal insulation of waterborne acrylic coatings through controlled integration of hydrophobic silica aerogel: Experimental insights and advanced modeling

  Journal of Materials Research and Technology · 2025-09-01 · 2 citations

  articleOpen access

  This research introduces a thermal insulation coating composed of hydrophobic silica aerogel dispersed in waterborne acrylic resin. To preserve aerogel's insulating efficiency, a waterborne resin was employed to limit resin intrusion into aerogel pores. Experimental evaluation of thermal conductivity using thermal resistance superposition demonstrated a significant decrease in thermal conductivity, achieving up to 46 % reduction at 50 vol% aerogel loading. Analytical modeling using Maxwell and Bruggeman models, alongside finite element simulations, revealed that the observed experimental performance was impaired by partial resin intrusion into aerogel pores, confirmed through thermogravimetric analysis (TGA). The intrusion depth and its impact on insulation efficiency were quantitatively modeled using a core-shell approach, providing a foundation for future optimization strategies.

  PublisherDOI

• Optimizing Thermal Insulation of Waterborne Acrylic Coatings Through Controlled Integration of Hydrophobic Silica Aerogel: Experimental Insights and Advanced Modeling

  SSRN Electronic Journal · 2025-01-01

  preprintOpen access
  PublisherDOI

• Integrated sizing and management of residential energy systems for electric and hydrogen vehicle charging

  Journal of Building Engineering · 2025-11-20

  article
  PublisherDOI

• A Vision-Language Model Agent for building code compliance

  2025-11-11

  articleOpen access

  Traditional methods for ensuring building code compliance often demand substantial time and resources and are prone to human error, leading to inconsistent evaluations of critical residential systems. Such inconsistencies can result in overlooked safety hazards and costly future repairs. To address these challenges, this paper introduces an innovative Vision-Language Model (VLM) agent specifically designed for building code compliance. The proposed agent combines advanced reasoning and action capabilities with specialized tools. It leverages a knowledge base comprising key building codes, including the International Residential Code (IRC) and the International Plumbing Code (IPC), and employs Retrieval-Augmented Generation (RAG) to identify relevant standards tailored to specific compliance requirements. An interactive interface enables users to submit both images and text, which the agent systematically analyzes. The VLM agent detects critical components, such as P-traps, and retrieves corresponding building code references. The system then generates a comprehensive report summarizing identified issues, assessing their severity, and citing relevant code sections. We use four distinct building components from real home inspection reports to evaluate the system's performance. The VLM agent achieves an average 96.25% similarity with the human-created inspection report. This research demonstrates a practical application of VLM agents, significantly enhancing the accuracy, accessibility, and reliability of building code compliance processes.

  PublisherOA PDFDOI

• Navigating the Social-Emotional Landscape of Neurodiversity in AI Education

  2025-08-21

  article
  PublisherDOI

• A graph attention network framework for generalized-horizon multi-plant solar power generation forecasting using heterogeneous data

  Renewable Energy · 2025-02-03 · 18 citations

  article
  PublisherDOI

• Preparing Autistic Students for the AI Workforce

  2025-06-23

  articleOpen access

  Software project courses too often focus instruction on technical skills but leave necessary communication and teamwork skills as an exercise for the learner. When autistic students take these courses, they often find difficulty fitting into teams with non-autistic students because of their different styles of communication. To explore how to help autistic students adapt to non-autistic normative software engineering environments, we designed and taught a first-of-its-kind online project course on AI, explicitly teaching communication and teamwork skills with purposefully designed scaffolds. After the course, students were matched with professional, online summer internships in which they could apply the skills they learned. We detail the course structure, including the pedagogical strategies employed and the specific challenges encountered. Our experiences reveal key elements that contributed to the course's success, such as the importance of adaptive teaching methods and the need for carefully considered instructor training for teaching neurodivergent learners in software engineering. We share this report to provide guidance for educators, researchers, and advocates seeking to develop effective computing education programs that include autistic students.

  PublisherDOI

Frequent coauthors

• Behnam Mohammadi‐Ivatloo

  Lappeenranta-Lahti University of Technology

  81 shared

• Morteza Nazari‐Heris

  Lawrence Technological University

  38 shared

• Ebrahim Karan

  Sam Houston State University

  38 shared

• Atefeh Mohammadpour

  California State University, Sacramento

  32 shared

• Mohammadreza Daneshvar

  University of Tabriz

  30 shared

• Marwa Hassan

  Louisiana State University

  27 shared

• Yewande S. Abraham

  26 shared

• Saratu Terreno

  Bradley University

  24 shared

Education

• M.S., Architectural Engineering

  Pennsylvania State University