Humanoid Robots and Autistic Children: A Review on Technological Tools to Assess Social Attention and Engagement

Abstract

Recent studies suggest that robot-based interventions are potentially effective in diagnosis and therapy of autism spectrum disorder (ASD), demonstrating that robots can improve the engagement abilities and attention in autistic children. While methodological approaches vary significantly in these studies and are not unified yet, researchers often develop similar solutions based on similar conceptual and practical premises. We systematically review the latest robot-intervention techniques in ASD research (18 research papers), comparing multiple dimensions of technological and experimental implementation. In particular, we focus on sensor-based assessment systems for automated and unbiased quantitative assessments of children’s engagement and attention fluctuations during interaction with robots. We examine related technologies, experimental and methodological setups, and the empirical investigations they support. We aim to assess the strengths and limitations of such approaches in a diagnostic context and to evaluate their potential in increasing our knowledge of autism and in supporting the development of social skills and attentional dispositions in ASD children. Using our acquired results from the overview, we propose a set of social cues and interaction techniques that can be thought to be most beneficial in robot-related autism intervention.