Real-Time Guidance for Adults with Autism Spectrum Disorder (ASD) using Sensor Fusion, AI and Augmented Reality

Project Description

High-functioning young adults with autism spectrum disorder (ASD) have high unemployment rates because of a range of social interaction and communication deficits, despite often having college degrees, average to high IQs, and various useful technology skills. Currently, young adults with ASD and their families, along with companies that could benefit from employing them, are inadequately equipped with training and coaching modalities that can facilitate the transition to work. Most computer-based assistive technologies for ASD focus on children, and on low and medium-functioning individuals, given the devastating effects of severe impairments and the importance of early intervention. Relatively few programs aim at high functioning ASD for young adults transitioning to work. Effective training for this population has great potential for rapid huge economic benefits, due to the large number of individuals with ASD aging into adulthood each year, and the impact of bringing each such person to long-term full-time employment.

The goal of this project is to create multiple systems to improve workplace efficacy of young adults with ASD including a range of workplace communication trainings tools, such as video games, role-playing scenarios, and technology-enabled training systems. The training systems are based on virtual reality (VR) and augmented reality (AR) that will enable high functioning individuals with ASD to develop workplace-appropriate behaviors, such as looking at conversational partners, maintaining interpersonal distance, orienting the body for dyadic and triadic conversations, and providing appropriately timed engagement responses when listening. The key advantages of VR/AR systems compared to human coaching for ASD include repetition, structured progression, attractive gamification, lower cost, reassuring controlled environments, and immediate real-time feedback. To overcome the problem of generalization to real-world situations, coaching will extend from VR to the related AR systems, allowing behaviors to be practiced with real people.

In addition to VR/AR head-mounted systems, light detection and ranging (LiDAR) systems will be used for monitoring of the workplace environment. Such external monitoring can address a wider range of behaviors, such body orientation, gestures, and posture (for example, a person should not have their head down on a desk during a work meeting). The LiDAR system enables us to calculate interpersonal distances precisely and to decide whether the body orientation is appropriate or not, with the help of various algorithms and machine learning models. Unlike a camera, a LiDAR system preserves privacy.

The project will build upon a successful summer internship program held by Dr. Chukoskie in the summer of 2018, in which 25 young adults with ASD worked to design and develop healthcare-related video games under the guidance of coaches. After the outstanding feedback from participants and family members, the internship program is going to continue in future summers, with the addition of new automated coaching methods and systems each summer. The impact of this project is potentially huge, as 500,000 individuals with ASD reach working age over the next decade. Furthermore, this application which intersects health and engineering will attract women and under-represented minorities to careers in STEM disciplines, allowing the faculty to amplify their outstanding track records on mentoring and outreach. Successful modeling of virtual coaching can be extended to other behavioral coaching needs across various healthcare applications, for coaching of healthcare providers as well as patients.


Principal Investigator
Co-Principal Investigator
Sujit Dey
Craig Callender
Shana Cohen
Joshua Shapiro
Research Students
Wenchuan Wei
Onur Tepencelik
Saygin Artiran
Sundar Rengarajan
Trent Simmons
Jada Wiggleton-Little
Jessica Miguel