Study of efficacy of applied behavioral analysis-based (ABA) robotic system in early intervention training of autism spectrum disorder (ASD) children

Abstract

Autism Spectrum Disorder (ASD) in children can be observed by difficulties with social, executive dysfunction, communication and sensory issues. Although there are various well-established interventions for ASD children, most of them are done manually by human therapists and parents which require attention to details and time-consuming. The objective of this research is to develop Human-Robot Interaction (HRI) platform using partial application of Applied Behavioral Analysis (ABA) techniques for early intervention programmes of ASD children. The ABA techniques were then converted to Finite State Machine (FSM) and were programmed in Python 3 (Linux environment) to enable future researchers for HRI to implement the correct technique in lesser time. This framework is then evaluated on ASDC. Four interaction modules with predictable and consistent teaching structures were developed. For implementing the robot modules in this work, the FSM approach were used during the integration of all three components of interaction modules: (a)learning content (developed with ABA therapist); (b)rigid teaching structures; (c)robot’s response. Interaction modules, proof-of-concept and final framework validation were done by experienced ABA therapist. The modules were converted to FA where it enables future researchers to iterate the correct method more effectively and to lessen the time-consumed in developing HRI for ASDC. Efficiency of the system were tested with 20 ASD children recruited from IDEAS. 95% of the subjects managed to finish the modules. From the 19 subjects, 53% subjects showed excitement over meeting the robot, and the remaining 47% showed no expression. While they showed indifference over meeting the robot, their interest started to peak once reinforcement (video reward - cartoons) were shown on the monitor. Out of the 19 subjects, only 16% showed excitement throughout the session while the others started to get restless by the end of the goodbye module. The last subject (9 years old boy, male) was not able to finish the modules as he showed clear signs of irritability. While the subjects showed varying levels of emotions during the interaction, and some were not able to finish the modules, the preliminary results obtained can contribute to better understand factors that might help determine sub-groups of children with ASD for whom robots could be particularly useful, and the structure they need. In summary, two objectives involving development of HRI framework for ASDC were fully achieved while a mixed results on effectiveness of the system were obtained when evaluating efficiency of the system.