Abstract:

This work presents a collaborative learning methodology designed to strengthen students’ ability to analyse, formalise and program the automation of industrial systems in small groups. The approach has been applied in an “Industrial Automation” course designed for Electrical Engineering students at our university, where groups work collectively to interpret real-world process descriptions, negotiate meanings, identify ambiguities in natural-language specifications and jointly construct GRAFCET diagrams, input/output lists and PLC programs. The collaborative dimension is central to the methodology: each team must discuss, refine and agree upon the functional requirements before progressing to implementation, encouraging critical thinking, shared decision-making and deeper understanding of the automation tasks.

To complement this collaborative framework, the experience integrates a conversational AI agent that simulates a non-technical plant manager. This agent provides informal, incomplete and sometimes ambiguous descriptions of the production system, requiring each student group to formulate questions together, compare interpretations and reach consensus on the system’s behaviour. Although the AI component is not the core of the methodology, it adds a novel interactive layer that enhances realism, increases engagement and supports the development of key competencies related to requirement gathering and clarification.

The methodology has been implemented with different simulated industrial systems and with varying profiles of simulated plant managers, allowing students to experience distinct communication styles and levels of clarity. Quantitative data show improvements in learners’ performance in collaborative tasks and in the quality of their automation designs, while qualitative feedback highlights increased motivation, a stronger sense of realism and better team coordination. In addition, the paper presents results from similar collaborative methodologies implemented in the final year of secondary education, and introduces planned adaptations in which the intelligent agent will act as a virtual client that students interview to clarify how an industrial system should work, as a tool that scores the depth and quality of their questions, and as a simple tutor that guides them while programming the system in C and C++, following the same methodology used at the university level.

The paper describes the collaborative design of the activity, the role of the AI agent as a complementary innovation and the learning outcomes observed, offering practical guidelines for the integration of group-based problem-solving and conversational AI simulations in engineering and pre-university education.

Keywords:

Collaborative learning, Conversational AI in education, Engineering education, AI-supported programming learning, Secondary–university transition.