Safe and Adaptive Social Robots for Children with Autism

Abstract

Social robots are being considered to be a part of the therapy for children with autism due to the reported efficacy such technology in improving the outcomes. How ever, children diagnosed with autism exhibit challenging behaviors that could cause harm to themselves and to others around them. Throwing, hitting, kicking, and self harming are some examples of the challenging behaviors that were reported to occur among this population. The occurrence of such behaviors during the presence of a social robot could raise some safety concerns. For this reason, this research attempts toidentify the potential for harm due to the diffusion of social robots and investigate means to mitigate them. Considering the advancement in technology and the progress made in many computer science disciplines are making small and adaptable social robots a foreseeable possibility, the studies presented here focus on small robotic form factors.The first study quantities the potential harm to the head due to one of the identi?ed risky scenarios that might occur between a child and a social robot. The results re leaved that the overall harm levels based on the selected severity indices are relatively low compared to their respective thresholds. However, the investigation of harm due to throwing of a small social robot to the head revealed that it could potentially causet issue injuries, sub-concussive or even concussive events in extreme cases. The second two studies are aimed to make small robots safer by optimizing their design. Hence,studies are conducted investigating how robot design can be made safer by investigating different design factors. The study investigated the in?uence of the mass and shape on the linear acceleration of a developed dummy head. The results revealed that the two design factors considered (i.e. mass and shape) affected the resultant response. The second study investigated the in offence three different soft material sonthesa meresponse. The endings showed that the control factors considered are not statistically significant in attenuating the response. Finally, the last two studies attempt to make small robots more adaptable to promote safer interactions. This is carried out by em bedding the recognition of unwanted physical interactions into companion robot with the appropriate timing of responses. The findings of the first study highlight the pos sibility of characterizing children's negative interactions with robotic toys relying on accelerometer sensor. The second study showed that producing a late response to an action (i.e. greater than 1.0 s) could negatively affect the children's comprehension of the intended message. The work presented in this dissertation is multidisciplinary that involves the field of Mechanical Engineering and Information Technology.