Developing multimodal perception module for autism diagnosis

Abstract

Early diagnosis of autism and reception of intervention at key developmental stages of the child are found to have long-term e ects in development of social and cognitive skills in Children with Autism (CWA). Lack of biological markers and complexity of diagnostic methods make the diagnosis highly subjective to the experience and perceptions of the specialist performing the diagnosis. There is a dire need for identifying technology assisted solutions to meet the needs of autistic community owing to the lack of su cient infrastructure and mental health professionals and also due to the high subjectivity in the diagnosis attributed by the intricacies of diagnostic procedure. Studies have shown that the capabilities of commercially available robots could be leveraged in the development of robot-assisted interventions. In India, there has been limited research focusing on the use of assistive technologies for designing interventions for CWA. With the help of this project, we aim to create a multi modal autism diagnosis tool which will provide detailed and regular observations to the therapist regarding di erent actions of the child. We leverage the use of the robot NAO to conduct di erent tasks like Response to Name Call, Joint Attention and Free Play.