Engagement of Children with ASD Using a Tactile Robot

Friday, May 12, 2017: 10:00 AM-1:40 PM
Golden Gate Ballroom (Marriott Marquis Hotel)
R. Chukoskie1, T. S. Chou2, L. Chukoskie1, J. Krichmar2 and J. Townsend1, (1)University of California, San Diego, La Jolla, CA, (2)Cognitive Science, University of California, Irvine, Irvine, CA
Background: Individuals with autism spectrum disorder (ASD) have widely recognized challenges in social communication. Social robots have been designed and used to provide an alternative and perhaps easier way to promote engagement and communicative exchange in children with ASD. Here, we characterize the behavioral effects of children and adolescents with ASD using a tactile robot, not explicitly designed to promote engagement or communication. CARBO (short for CARetaker RObot) was built by incorporating a spiking judgment neural network robot “brain” (Chou, et al., 2015) that has the potential to learn from interactions. The robot’s shell is covered with trackballs creating a smooth surface for tactile interaction. Two games (ColorMe and FollowMe) request hand movements in a particular pattern on the shell to illuminate LEDs beneath each trackball. Eight directions of the trackball are mapped to different colors of the underlying LEDs. The games are designed engage participants in producing a wide range of movements, as well as turn-taking activities that paint CARBO’s shell.

Objectives: Through a series of structured play sessions with CARBO and another interactive toy serving as a comparison, we aimed to objectively characterize behaviors during interactions. The results of these interactions will be used to improve the games and to create a version of CARBO that can serve as a bridge in behavioral therapy for children with ASD.

Methods: We recorded multiple sessions of children and adolescents interacting with a tactile robot to test if interactions with CARBO modified sensory and motor behaviors and/or engagement and communicative behaviors in individuals with ASD. Using the data from CARBO, we assessed changes in movement behavior during game play. Using video data, we employed a dictionary of behaviors (including gestures and utterances) that we created for video-based coding and compared behaviors observed while each participant interacted with CARBO and another interactive toy. We also examined how behaviors changed over multiple visits. All coding was conducted using the multimodal annotation tool ELAN.

Results: We characterized interactions in 6 children and adolescents with ASD. These individuals spanned the continuum from non-verbal to very facile with verbal communication, and ages 9-17. Participants exhibited longer duration engaged interactions with CARBO than with other toys. We also noticed fewer instances of repetitive behaviors while participants were engaging with CARBO. Responses to CARBO’s requests to change movement direction or speed were responded to promptly in most cases.

Conclusions:  Our pilot study observations suggest that children are considerably more engaged with the tactile robot, CARBO than with an interactive game. The extended duration of the engagement and ability to use the robot in requesting and turn-taking behaviors make it a useful tool in behavioral therapy.