Temi Dice: A cognitive-motor game integrating social robotics and computer vision

Abstract

This paper presents Temi Dice, a playful application installed on the Temi social robot and designed as a cognitive-motor activity based on the traditional game “Simon Says.” The system provides verbal instructions to the user, who must perform a gesture only when the command is preceded by the phrase “Temi says.” This dynamic allows for the development of skills related to attention, inhibitory control, and motor coordination. The user’s responses are evaluated using computer vision with MediaPipe, which extracts body points and applies defined geometric rules. The system recognizes gestures without the need to train a specific classification model. Additionally, the robot provides real-time verbal feedback during the game. Usability was assessed using the SUS questionnaire, yielding a high average score. The results suggest good initial acceptance and highlight the potential of social robotics and computer vision for designing intuitive stimulation activities for older adults.