Stress Detection of Children with Autism using Physiological Signals in Kaspar Robot-Based Intervention Studies

This study aims to develop a stress detection system using the blood volume pulse (BVP) signals of children with Autism Spectrum Disorder (ASD) during robot-based intervention. This study presents the heart rate variability (HRV) analysis method to detect the stress, where HRV features are extracted from raw BVP signals recorded from an E4 wristband during interaction studies with the social robot Kaspar. Low frequency power (LF) and high frequency power (HF) features are analyzed, and the results are verified with facial emotion analysis of the children with ASD. 21 children from 3 countries participated in the study. The results showed that physiological signals combined with affective state labels may predict the stress of children, and the children were not stressed overall their interaction with the Kaspar robot. In specific cases, the children started their session as stressed but their stress declined by the end of the session. These findings are also supported by the results of the vision-based affective state analysis.