BUILT2AFFORD, an initiative developed by Notre Dame faculty, aims to help low-income households in South Bend save energy. The program's goal is to develop and test technology that can be used to retrofit older houses in the South Bend community in order to save energy and reduce costs, which is how the project gets its name.
The project, led by three Notre Dame faculty members, Ming Hu, associate dean for research at the School of Architecture, Chaoli Wang, professor of computer science and engineering and Matthew Sisk, associate professor of the practice in the Lucy Family Institute for Data & Society, uses Google Street View to complete a detailed analysis remotely, on a large scale and in a non-intrusive way.
The National Science Foundation issued a grant to Notre Dame that will assist the researchers in developing this technology.
In the South Bend area, many of the older homes are less energy efficient, which comes at a great expense to their residents. Many homes are not adequately insulated or lack properly sealed windows, causing the energy that is spent heating and cooling these houses to seep through windows and insulation.
Another issue the project hopes to address is the presence of lead, which is present in many South Bend homes built before 1970.
“One of the effective ways we conceal the lead paint is to paint over it, but imagine the house is too humid. The paint will peel down, so the lead will be exposed,” Hu said.
The team is developing a dashboard that helps South Bend residents identify simple renovations they can make to save energy, improve air quality and cut costs. Their current research includes analyzing existing houses in South Bend and simulating how much energy can be saved if they undergo a certain renovation.
“As a building scientist, I can extract the thermal value from the building material and determine issues based on the year of the build, based on the material and considering the construction time. That is my expertise, and Professor Wang’s expertise is to build a computer vision model to read the material,” Hu said.
The researchers' process begins by analyzing a building from its Google Street View images. Then, they identify cost-saving solutions and train the technology to be able to recognize the same issues.
Residents will be able to evaluate the best way to save energy in their own homes remotely.
“We want to create a tool that everyone can use," Hu said.
The initiative is currently in the first of two stages. The first stage is to develop and test the technology, while the second stage involves validating their research by physically retrofitting houses in South Bend.
“The retrofit solution is pretty straightforward,” Hu said. “It is nothing new, exactly. Our technology aims to identify the house, to figure out the exact window or insulation level we need to replace is the research part.”
She continued, “We work with both Ph.D. students and graduate students along with some undergraduate students to help us do field auditing. Most of them right now are doing the Google Street View image processing.”
There are many undergraduate research assistants participating in research for the project, including freshman Andres Perez.
“Last semester, I assisted with fieldwork, visiting homes associated with the project to
capture thermal images of potential leakage points–such as doors, windows, ceilings–to identify possible energy leaks. These areas, especially in older houses, could lead to
higher energy costs and health concerns in colder climates for more susceptible groups. Earlier in the project, I also worked on annotating image data relating to windows,” Perez wrote in a statement.
Currently, Perez is assisting a graduate computer science student, working to train a machine learning model to assess the condition and features of houses using image datasets of cities.
“The CS undergraduate students I know, including myself, are focused on downloading
the previously mentioned image data. We are expanding our scope to multiple states,
which requires a substantial amount of data to accurately account for the diversity of
housing conditions,” Perez wrote.
The BUILT2AFFORD project has recently submitted its second stage proposal.
“We proposed that we retrofit eight houses in South Bend, and we record three months of data and use that data to validate the tool. And after that, we push it out so everyone can use the data, and the city will adopt the tool as well,” Hu said.
Siyuan Yao, a fourth-year Ph.D. student, shared his overall experience with the BUILT2AFFORD research.
“This project has given me the opportunity to collaborate with a large group of dedicated students. Both the professors and students involved are highly enthusiastic and hardworking. I am grateful to be part of such an energetic and collaborative team,” Yao wrote in a statement.
Yao said that the project has also broadened his understanding of housing concerns in South Bend.
“My hope is that we can eventually build a nationwide database that will not only benefit our own research but also be valuable for others working on related projects,” he wrote.
