Skip to main content

Robotics and Autonomous Systems Lab
at Vanderbilt

Intelligent Agent

Personnel: Lian Zhang, Ashwaq Zaini Binti Amat, Chuang K. Tang, Amy Swanson, Amy Weitlauf, Zachary Warren, and Nilanjan Sarkar.

Goals/Objectives:

  1. Effectively measure important aspects of dynamic interactions within a collaborative virtual environment
  2. Realize strategies for altering the environment to improve these aspects of interaction

Outline:

A Collaborative Virtual Environment (CVE), which is a computer-based and distributed virtual space for multiple users to interact with each other, may have the advantage to support flexible, safe, and peer-based interactions for ASD intervention. One challenge of using CVEs for ASD intervention is to evaluate the peer-based interactions. An intelligent agent that can communicate and interact with a child with ASD in a specific CVE may provide a way to address this challenge. In this project, we developed an intelligent agent that can communicate and play collaborative puzzle games with a child with ASD in order to evaluate the communication and collaboration skills of the child in a specific CVE. The intelligent agent was developed with a hybrid method, which combines a dialogue act classifier and a finite state machine. This hybrid method enables the agent not only to communicate and play the games but also to generate related features for the evaluation. A preliminary study with five children with ASD was conducted, and its results demonstrated the agent’s capability to communicate and play games with a child as well as the potential to evaluate his/her communication and collaboration skills in the CVE. The agent will be used to evaluate the skills of children with ASD when interact with their TD peers in the CVE in the future.