<p>Health education emphasizes the public’s understanding, internalization, and practical application of health-related behaviors. Using primary school students as an example, health education that lacks age-appropriate and engaging instructional design may result in reduced learning motivation within this group. To date, health education has predominantly targeted caregivers, thereby limiting opportunities for primary school students to develop the skills to manage their health independently. Given that primary school students benefit from concrete aids to support logical thinking, this study proposed a robot-assisted multimodal learning model grounded in the ARCS (Attention, Relevance, Confidence, Satisfaction) framework. To evaluate its effectiveness, an experiment related to asthma health education was conducted involving 60 primary school students aged 6 to 12&#xa0;years from a medical center in northern Taiwan. Participants were randomly assigned to either the experimental group, which adopted the robot-based ARCS (R-ARCS) learning model (<i>n</i> = 30), or the control group, which used the conventional ARCS (C-ARCS) learning model (<i>n</i> = 30). Results indicated that, compared to the C-ARCS learning model, the R-ARCS learning model more effectively enhanced primary school students’ learning achievement and motivation, while also reducing their asthma anxiety. Interview results further revealed that the experimental group experienced increased learning motivation, improved comprehension, and reduced anxiety. The findings provide a valuable reference for designing health education curricula targeting primary school students. Future research is recommended to incorporate parent–child collaborative learning models to optimize curriculum development and further enhance learning effectiveness.</p>