How are you feeling? Design and testing emotional dictionary for NAO robot to be used with children

This paper presents the development and preliminary evaluation of an emotional dictionary for the NAO humanoid robot, to be used in games activities with children with Autism Spectrum Disorder (ASD). The emotional dictionary integrates multimodal elements, such as gestures, LED eye colours, and speech, to enable the robot to convey positive and negative emotional states during interactive game scenarios. These scenarios are designed to support the development of social and cognitive skills in children with ASD. To create the emotional dictionary, eight gestures (four positive and four negative) were identified from existing literature and refined through an iterative co-design process involving psychologists, therapists, and engineers. The emotional expressions were further enhanced using eye colour changes (yellow/orange for positive, blue/violet for negative) and context-specific verbal feedback. A preliminary evaluation was conducted with 34 healthy young adults to validate the ability of the robot to correctly elicit emotions. Participants rated the robot’s emotional expressions based on valence and identified in emotional categories. The results demonstrated high recognition accuracy for both positive (e.g., happiness) and negative (e.g., sadness) combinations, with valence ratings distinctly separating these emotional states. These preliminary results provide guidelines on which combinations are most suitable to be used as feedback in games activities. Future work will incorporate these combinations into interactive game dynamics and extend testing to both typical and atypical children.