Teachy Mini: Development and Preliminary Evaluation of a Knowledge-Based Generative Social Robot for Higher Education

Synthesis: Vonschallen, Kaufmann, Oberle, Eyssel, and Schmiedel (2026) operationalize knowledge-based design (KBD) requirements for generative social robots (GSRs) by implementing them in the Reachy Mini robot platform through system prompting, retrieval-augmented generation, and stateful prompt orchestration, producing Teachy Mini — a GSR tutoring system for higher education. Recognizing that GSRs powered by LLMs offer personalized tutoring but also risks (misinformation, missing transparency, reinforcing incorrect responses), the authors conducted a preliminary evaluation study in which participants (N = 24) completed a robot-guided learning session about research methodologies, learning with either the robot or another mode.

Key Findings

Generative social robots (GSRs) powered by LLMs offer new possibilities for personalized tutoring in higher education but introduce risks related to misinformation, missing transparency, and reinforcing incorrect student responses.

Prior work identified knowledge-based design (KBD) requirements defining the informational prerequisites for GSRs to manifest responsible and effective tutoring behaviour.

The authors operationalized selected KBD requirements in the Reachy Mini robot platform through system prompting, retrieval-augmented generation, and stateful prompt orchestration, producing Teachy Mini.

A preliminary evaluation study (N = 24) had participants complete a robot-guided learning session about research methodologies, learning with either Teachy Mini or an alternative mode.

Study Design & Method

This is a system development and preliminary evaluation study. The researchers operationalized knowledge-based design requirements for generative social robots by implementing them on the Reachy Mini robot platform using system prompting, retrieval-augmented generation (RAG), and stateful prompt orchestration, building the Teachy Mini tutoring system. A preliminary evaluation study with 24 participants had them complete a robot-guided learning session about research methodologies, comparing learning with Teachy Mini against an alternative mode to assess the system's effectiveness and the value of the KBD approach.

Implications for AI in Education

Teachy Mini demonstrates how knowledge-based design requirements can be concretely implemented in generative social robots for Higher Ed tutoring, using system prompting, RAG, and stateful prompt orchestration to mitigate risks such as misinformation and reinforcing incorrect responses. It connects to Generative AI, large language models, tutoring, and Educational Robotics. For designers and educators, it provides a validated example of translating responsible-AI design principles into a functioning embodied tutor, complementing the prior KBD requirements study.

Limitations

The evaluation is preliminary with a small sample (N = 24), and the specific comparative learning outcomes are not fully detailed in the abstract. The system is built on the Reachy Mini platform, so generalizability to other robot platforms and to broader disciplinary content requires further study. The focus is on research-methodology tutoring, and longer-term effects and broader responsible-AI risks warrant additional evaluation.

Connected Concepts

Connected Articles

Knowledge Based Design Generative Social Robots 2026 — Knowledge-Based Design Requirements for GSRs

Task Context Trust Educational Hri 2026 — Task Context and Trust in Educational HRI

White Wu Robotics AI Education 2026 — Robotics and AI in Education

Citation

Vonschallen, S., Kaufmann, K., Oberle, D., Eyssel, F., & Schmiedel, T. (2026). Teachy Mini: Development and preliminary evaluation of a knowledge-based generative social robot for higher education. arXiv:2607.22345.