Artificial intelligence analysis of the emotions of autistic students in social interaction with the NAO robot

Abstract

Currently, technology is the most widely used tool in the development of daily life activities. The number of fields of knowledge that benefit from its versatility and application in the development of their activities is increasing. In the educational environment, it allows the generation of activities adapted to the needs of students. In recent years, robotics and artificial intelligence are the most widespread. The characteristics of these tools favour their application with students with autism spectrum disorder. Therefore, the objective of the research is to explore the integrated use of robotics and AI to promote communication and social interaction in students with autism, analysing the emotions they show in the designed activities. For this purpose, a pilot study was implemented with the NAO robot and four autistic children who developed imitation, game and social interaction activities. An automatic system based on convolutional neural networks was used to detect mood states in the interaction process. The results show that sadness, happiness and anger are the emotions most likely to occur in the participants. Therefore, it is concluded that the robot and the artificial intelligence system are a fundamental element to help express emotions in social interaction.