Personalized Affective Robotic Assistive Technology for Children with Autism

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Social robots are gradually becoming an integral part of human livelihood and have achieved significant results in healthcare, education and entertainment. Recently, research has geared towards identifying ways that such robots can be harnessed in special needs education for children with autism spectrum disorder (ASD). Autistic children have deficits in social interaction, communication and often portray repetitive behaviours. Although technology-based intervention strategies for autistic children could promise great results, many autistic children from resource-constrained environments have been left behind due to the cost implications and technical requirements associated with robotassisted learning. This thesis focused on investigating the suitability of a humanoid robot as an assistive technology for Ghanaian children with autism and proposing strategies for personalization of robot-mediated learning sessions. An iterative prototyping approach was used to design and develop a novel low cost humanoid robot, RoCA, which has been used together with another robot Rosye in a series of empirical studies to assess the initial reaction of some autistic children towards the robots and the engagement levels of the children through longitudinal studies. Results from the preliminary and longitudinal studies indicated that the robot was able to engage some of the children in imitation and general tasks and also succeeded in persuading some children to perform the robot’s requests via multiple prompted cueing. The thesis also presents a deep fuzzy framework for personalized affective robot assisted learning in autistic child￾robot interactions. This framework is based on a proposed deep learning model, SingleShot Emotion Detector (SED) and a fuzzy based engagement prediction engine ii which can use scores, IQ levels and task difficulty as input variables for estimating the engagement levels of autistic children while interacting with social robots. The framework was implemented in the humanoid robot RoCA and another empirical study was conducted to evaluate the effects of the personalization approach provided by the deep fuzzy framework on learning gains in autistic child – robot interactions. Statistical significance of improved learning gains associated with the deep fuzzy approach adopted by RoCA was confirmed by Mann Whitney tests. The thesis also investigated the behavioural intention of special needs teachers to use robots in the classroom to teach autistic children using Unified Theory of Acceptance and Use of Technology (UTAUT) as research model. The results indicated that performance expectancy, effort expectancy and social influence positively affect the behavioural intention of special needs teachers to use robots to teach children on the autism spectrum.
A thesis submitted to the Department of Computer Science, Faculty of Computational and Applied Sciences, College of Science, in partial fulfilment of the requirements for the degree of Doctor of Philosophy.