Exploring effectiveness of robot-assisted instruction to teaching social engagement of school children with autism

Background. Children with autism spectrum disorder (ASD) demonstrates deficits in social communication and social interaction across contexts. The educational use of social or interactive robots is promising in enhancing social engagement of children with ASD because children with ASD exhibit strengths in understanding the object-related world over the social world; they are more responsive to feedback when administered via technology rather than a human; and they are intrinsically interested in activity when it involves electronic or robotic components. However, evidence-base of robot-assisted educational programs for children with ASD is limited. This study examined the effectiveness of using robotic educational program to enhance social engagement amongst children with ASD, specifically, the frequency and duration of eye contact, and verbal initiation of children with the human instructor.
Methods. A quasi-experimental, ABA time-series design was employed in this study to examine how the effects of robotic interventions were manifested over time. 15 children aged 5 to 11 attending special and mainstream schools, who had been diagnosed with ASD were recruited to this study.
Participants enrolled in the intervention program had received a total of 12 sessions on social skills training. These twelve sessions were structured into a baseline phase (3 weeks), an intervention phase (6 weeks) and a generalization phase (3 weeks). The baseline phase was acted as a control on the experimental condition. A series of baseline measures were taken in the baseline phase (3 weeks), in which the children received social skills training only by the human instructor. In the intervention phase (6 weeks), the robot was integrated into the training program to assist the instructor to conduct the structured interactive social game, structured story-based activities and structured singing/ dancing activities. In the generalization phase (3 weeks), the session was switched back to the original training mode without the participation of a robot.
Measurements were taken on a weekly interval in all three phases. Interactions of the child with the instructor in the session were video-recorded. A video coding procedure was performed to measure the three target behaviours – frequency of eye contact, duration of eye contact and frequency of verbal initiation. Time-series data on the three variables were analysed using graphic visual analysis and a non-parametric approach called Tau-U.
Results. Results from both the visual analysis and Tau-U analysis had confirmed the robotic educational program were effective in improving the eye contact (both frequency and duration) and verbal initiation of children with the human instructor. The improvement of the children were sustained even with the withdrawal of the robot in the generalization phase.
Conclusion. This study confirmed a robot-assisted educational program was effective in enhancing the social engagement of children with ASD. Implications on practice were discussed and future development of using a social robot in educational programs was inspired through this project. Copyright © 2019 THEi AETI 2019.