Educational Integration of ESP32 Project-Based Learning with Generative AI: Design, Implementation, and Evaluation in Two Engineering Cohorts
Academic Article in Scopus
Microcontroller education is shifting from syntaxfocused programming to engineering reasoning under real constraints. We report a five-week, project-based and cooperative learning module using ESP32 with Arduino, where generative A I is allowed as guided scaffolding under mandatory traceability and individual oral verification. Two independent cohorts (n=14 and n=15; N=29) worked in triads with rotating roles and evidence boards. Summative assessment combined public prototype demonstration (50 percent) and individual viva (50 percent), while structured peer assessment supported accountability. The overall mean grade was 75.54 (SD 14.61); cohort means were 77.45 and 73.75, with a small between-cohort difference (Cohen d about 0.25). The approval rate was 72.4 percent. Peer assessment averaged 0.917 and correlated strongly with grades (r about 0.78 overall; 0.85 and 0.69 by cohort), indicating alignment between recognized contributions and summative performance. Student perception was high and homogeneous across teaching, guidance, environment, expertise, and challenge. We conclude that integrating traceable generative AI with product-viva-peer assessment can accelerate learning in short blocks without compromising validity, offering a replicable design for instrumentation and control courses. The study provides pedagogical insights into how GenAI-supported PBL can enhance engineering reasoning and accountability. It contributes an evidence-based design model that integrates authentic assessment, cooperative learning, and AI literacy for modern engineering education.
