AI aids those with ADD in construction

Body

Using virtual reality (VR) and artificial intelligence (AI), a team of researchers at George Mason University is taking a wrecking ball to barriers faced by neurodiverse individuals in construction.

The Mason team partnered with researchers at Purdue University on a $2 million National Science Foundation project. The project will explore the future of construction and identify ways to use human-robot teaming to open the field of construction to neurodiverse individuals, specifically those with attention deficit disorder (ADD) or attention deficit/hyperactivity disorder (ADHD).

People with ADD or ADHD have more variable attention than those without it, which includes great strengths and weaknesses. Their strengths include incredible problem-solving skills and creativity. Their tendency to be disorganized, have a wandering mind, and take part in more high-risk behaviors can limit their ability to succeed in specific fields, like construction. But the research team believes that with the help of new technology, we will better understand this condition.

“As the future of construction is moving toward technological coupling, workers and machines have to team up to accomplish project goals. Using AI, robots can learn to understand, react, and predict human behaviors. This collaboration could make it possible for us to bring neurodiverse people into construction,” says Behzad Esmaeili, assistant professor of civil, environmental, and infrastructure engineering, and the principal investigator (PI) of the Mason team.

Robots, or cobots, could learn to spot inattention, riskier behaviors, and mistakes while their human partners can be trained to work with these robots to improve their collaboration.

At least 5 percent of the population has been diagnosed with ADD and ADHD. Still, many are undiagnosed, says Brenda Bannan, professor of learning design and technology in the College of Education and Human Development and co-PI on the grant.

For AI robots to learn human behavior, they have to gather data, which has limitations on construction sites. “We cannot safely collect data from actual construction sites because it exposes people to potential injuries. Additionally, this is for a futuristic construction site, where humans and robots are working together,” says Esmaeili.

Craig Yu, assistant professor of computer science and co-PI, uses VR and works with the research team to develop a virtual environment to simulate different scenarios on construction sites, including risky and accidental situations. “We can run various simulations and collect plenty of data that AI robots and systems can learn from, without posing any risks,” says Yu.

The team can also use the virtual environment to train people and construction teams to work with these robots. “Using VR lowers costs and risks, and it can be used to further improve efficiency and productivity on construction sites,” says Yu.

Yu, Esmaeili, Bannan, and their co-PI Maurice Kugler, professor of public policy in the Schar School of Policy and Government, will employ the observations and data from numerous variable sensors, like eye trackers, cognitive brain monitoring, and other psychophysiological and biomechanical metrics to better understand the behavior of people with ADD and ADHD on construction sites and teach robots about how this population uniquely works.

“This innovative technology can give access to people with ADD to completely new employment opportunities and therefore substantially benefit workers, employers in the construction industry, and society at large,” added Kugler, who will conduct the cost-benefit analysis of the research project.

Their collaboration came from conversations at meetings for the Center for Advancing Human Machine Partnerships (CAHMP). “This is the purpose of CAHMP. It was designed to strike these transdisciplinary collaborations and projects,” says Bannan, one of the center’s founders.

The team has already screened nearly 300 participants, and they plan to start running VR simulations and collecting data in February.