Modeling and Analysis of Multimodal Collaborative Virtual Interaction for Autism Spectrum Disorder Intervention

Abstract

In recent years, computer-assisted systems have been increasingly employed to support a wide range of psychological activities from therapy to the training of specific skills. This dissertation describes my research on the design, development and application of Virtual Reality (VR) -based computer-assisted systems for Autism Spectrum Disorder (ASD) intervention. In addition to providing engaging and low-cost intervention environment, VR-based computer-assisted systems have the potential to create an alternative intervention paradigm to reduce resource burden and expand accessibility to effective ASD interventions. Previous studies on computer-assisted systems mainly focused on understanding and improving social communication and interaction deficits of children with ASD. Most of these studies used software-based restricted interactions between the user and the computer, and were not able to scale up well for complex adaptive interactions and thus led to weak transfer of learnt skills into real-world scenarios. In addition, few studies explored intervention of motor impairments of children with ASD using advanced computer technology. This is important since high prevalence of motor impairments have been found among ASD population and children with ASD show motor coordination deficits and delays in fine motor skills compared to their typically developing (TD) peers. The work presented in this dissertation was designed to address these issues and to provide more efficient and adaptive training systems for ASD intervention: (1) developing Collaborative Virtual Environment (CVE) systems that support realistic interactions and flexible conversations between remote users, (2) developing VR-based systems with haptic device that support the assessment and training of fine motor skills of users as well as the investigation of the impact of fine motor skill on the social skill of the users, and (3) developing an intelligent agent that supports human-to-agent communication and interaction within the VR-based fine motor training system. All these systems were validated with human subject studies involving children with ASD with encouraging results.