🧠 AI Ed Wiki

📄 Research Article

Would You Let a Humanoid Play Storytelling With Your Child? A Usability Study on LLM-Powered Narrative Human-Robot Interaction

◐ medium · 2026-08-13human-robot-interaction, social-robots, llm, storytelling, child-development, educational-robotics, usability

Synthesis: Lombardi et al. (2025) present a framework for enhancing the attention and social capability of the iCub humanoid robot by integrating advanced perceptual abilities that recognize social cues, understand surroundings through generative models such as ChatGPT, and respond with contextually appropriate social behaviour. They implement an interaction task using a narrative (storytelling) protocol in which a human and the robot co-create a short imaginary story, exchanging cubes with creative images. Experiments quantify the usability and quality of experience perceived by participants interacting with the LLM-powered narrative human-robot interaction system.

Key Findings

  • A key challenge in human-robot interaction is developing systems that can perceive and interpret social cues to enable natural, adaptive interactions.
  • The framework integrates generative models (e.g., ChatGPT) so the iCub humanoid can understand its surroundings and respond with contextually appropriate social behaviour.
  • The storytelling task has human and robot co-create an imaginary story by exchanging image-cubes, supporting collaborative narrative interaction.
  • The study validates the protocol and framework through usability and quality-of-experience measurement with participants.

    • Study Design & Method

    This is a usability/quality-of-experience study of an LLM-powered narrative human-robot interaction system built on the iCub humanoid platform. The researchers integrated perceptual capabilities for social-cue recognition with generative-model understanding (ChatGPT) and implemented a storytelling interaction protocol where the human and robot jointly create a story by exchanging cubes with creative images. Participants interacted with the system, and the degree of usability and quality of experience was quantified to validate the framework and protocol.

    Implications for AI in Education

    The work demonstrates how LLM-powered social robots can support collaborative, engaging learning interactions such as storytelling, relevant to Human Robot Interaction, Social Robots, and Educational Robotics. Integrating generative models enables robots to respond adaptively and contextually to children, supporting naturalistic educational interaction. This connects to large language models in education and to child-focused learning activities such as storytelling, with implications for how embodied AI agents can participate in co-creation and narrative learning.

    Limitations

    The study focuses on usability and perceived quality of experience rather than measured learning outcomes; sample sizes and context are not specified in the abstract. The iCub platform is research hardware with limited classroom availability, and the generalizability of the LLM-integrated interaction framework to other robot platforms and age groups warrants further study.

    Connected Concepts

  • Human Robot Interaction
  • Social Robots
  • Educational Robotics
  • LLM
  • Student Engagement

    • Connected Articles

  • Robobuddy LLM Social Robots Classroom 2025 — RoboBuddy: LLM-Powered Social Robots for Storytelling
  • Enhancing Creative Writing With Robot LLM Integration The Interplay Of Embodimen — Robot-LLM Integration and Embodiment in Creative Writing
  • Social Robot Study Companions — Social Robots as Study Companions

    • Citation

    Lombardi, M., Calabrese, C., Ghiglino, D., Foglino, C., De Tommaso, D., Da Lisca, G., Natale, L., & Wykowska, A. (2025). Would you let a humanoid play storytelling with your child? A usability study on LLM-powered narrative human-robot interaction. arXiv:2508.02505.