New Human-AI Teaming Workspace for Construction Workers with Neurodiversity

A project titled “FW-HTF-R: Collaborative Research: Worker-AI Teaming to Enable ADHD Workforce Participation in the Construction Industry of the Future” received financial support from the National Science Foundation.

Image Credit: 24Novembers/Shutterstock.com

The grant was awarded to Behzad Esmaeili, Assistant Professor, Civil, Environmental and Infrastructure Engineering; Lap-Fai (Craig) Yu, Associate Professor, Computer Science; Maurice Kugler, Professor, Public Policy; and Brenda Bannan, Professor, Learning Design and Technology, Division of Learning Technologies.

People suffering from attention-deficit or hyperactivity disorder (ADHD) have been disregarded in the construction work because of possibly greater risks of injuries. However, their special talents could be utilized with the help of an ecosystem of co-bots powered by artificial intelligence (AI).

Intelligent machines should have the ability to evaluate, adapt, and react to both workers and their surroundings for humans and machines to turn into true teammates—and correspondingly, for technology to offer occupational chances to people with such neurodiversity. Such agility needs a reciprocal teaming ability where workers can use their AI counterparts as more than tools, and AI systems could collaborate with workers smoothly by anticipating their behaviors.

For essential foundations to be laid for building this human-AI teaming workspace for construction workers with neurodiversity, this proof-of-principle project will translate noninvasive psychophysiological (for example, eye movements, brain signals, etc.) and biomechanical metrics (for example, gait, posture) into information that can be evaluated, modeled and leveraged by a personalized AI-based training system to anticipate behaviors of workers for enhanced worker-machine teaming.

For this aim to be achieved, step-by-step procedures will have to be followed in this project. The first one is to utilize mixed-reality simulated future job site settings and biomechanical or psychophysiological metrics to comprehend the technology interactions of ADHD workers at the time of human-machine collaborative tasks.

The second step is to develop and test AI algorithms to automatically capture interactions and performance and then to choose and offer optimal real-time feedback interventions on time to avoid injuries.

The third step is to examine the negative effects of wearable technologies and AI, such as privacy, integrity, security, usability, and other ethical problems inside teaming context.

The fourth step is to determine and quantify the social and economic effects of the suggested worker-AI teaming system to allow different workforce participation. The fifth step is to generate an AI-informed training teammate to encourage active, cooperative learning and to illustrate the possibility of human-AI teaming technology.

Data collection will occur in a multi-sensor immersive mixed-reality environment comprising of a virtual projection of the surrounding, passive haptics that mimic future construction sites, and environmental modalities that will capture their realistic reactions to their AI-teammates and analyze human performance efforts.

Furthermore, the gathered data will be utilized to design a multi­dimensional predictive model to avoid possible incidents. Also, this study will examine the planned work scenarios of worker-AI teaming, the undesirable impact of AI-teaming for workers, and the welfare of society.

Considering that 4.2 percent of workers are diagnosed with ADHD—a disorder that is associated with more than 120 million lost workdays in the USA each year, equating to a human capital value of $19.5 billion—this project’s efforts to enable diverse workforce participation in the construction industry will have positive social and economic impacts.

Behzad Esmaeili, Assistant Professor, Civil, Environmental, and Infrastructure Engineering, George Mason University

The scientists received financial support of $1,200,000 from NSF for this study. Funding started in October 2021 and will end in late September 2025.