Developing a 2-DOF robotic arm kit for embodied AI learning in middle school technology education

Abstract

Artificial intelligence (AI) literacy is increasingly emphasized in technology education, yet many middle-school classroom activities remain screen-based and offer limited opportunities to experience the full AI pipeline in an authentic design context. To address this gap, this study developed a low-cost, classroom-ready 2-degree-of-freedom (2-DOF) robotic arm kit and an eight-lesson AI-integrated instructional unit in which students assembled the robot arm, collected and labeled image data, trained and improved a classification model, and applied model outputs to robot-arm control. A quasi-experimental pretest–posttest control-group design was employed with 98 ninth-grade students in four intact classes (experimental group, n=46; comparison group, n=52). Quantitative data were analyzed using ANCOVA with pretest scores as covariates, and qualitative data from student reflection reports were analyzed thematically. The experimental group significantly outperformed the comparison group on value of AI, efficacy of AI, AI literacy, and total scores. Qualitative findings further showed that students recognized both benefits and risks of AI and proposed multi-layered safety strategies involving physical safeguards, operational rules, and data/model management. These findings suggest that embedding the full AI pipeline in a tangible engineering-design task can support embodied, responsibility-oriented AI learning in middle school technology education.