Towards a Structured Multimodal Speech-Image Coordination

摘要

This paper presents a novel approach to the use of the humanoid robot Pepper in educational contexts, focusing on dialogic and multimodal interaction for teaching abstract concepts in mathematics and physics.Unlike traditional lecture-based models, our system supports learner-centered lessons where students engage in spontaneous dialogue with the robot while interacting with visual content displayed on its tablet.The robot responds to user questions through the coordination of speech, synchronized textual output, and dynamic visual cues, such as real-time modifications to vectorial images that highlight relevant elements.To promote accessibility and engagement, the system includes customizable visual features such as color schemes and adjustable font settings for visual-impaired users.This approach aims to foster a more inclusive, personalized, and interactive learning experience by adapting the lesson content to the learner's interests and inquiries.