《Education and Information Technologies》2026年6月

Educational robots exhibit considerable potential for adaptive learning. However, current implementations often limit their role to knowledge transmission through teacher-centred approaches, potentially constraining pre-service teachers’ develop- ment of pedagogical skills. This study proposes and evaluates a knowledge graph- assisted educational robot (KGAER) system designed to scaffold structured domain knowledge and support self-regulated learning. Using a quasi-experimental design, pre-service teachers were assigned to an experimental group (n = 39) that learned with KGAER, or a control group (n = 46) that used a conventional educational ro- bot. Results indicated that the KGAER group demonstrated significantly higher instructional design quality (F = 11.91, p < 0.01, η² = 0.13), greater flow experience (t = 3.07, p < 0.05, d = 0.67), and stronger overall motivation (F = 8.02, p < 0.05) com- pared with the control group. Cognitive load did not differ significantly between groups, suggesting that the intervention did not impose an additional mental bur- den. These findings support the use of knowledge graphs to enhance the instruc- tional effectiveness of educational robots, illustrating how human–AI collaboration can maintain pedagogical agency while improving learning outcomes in teacher education.