Development and Empirical Study of a Multimodal Educational Agent System in Metaverse for Engineering Education Digital Transformation​

Abstract

Against the backdrop of digital transformation in higher education, engineering education faces critical challenges including the theory-practice divide and high barriers to technical tool adoption. This study develops a tripartite framework ("functional deconstruction-process modeling-data governance") for educational agent systems, which is built upon the Coze platform and rigorously aligned with the standards of China's Emerging Engineering Education (3E) initiative—for deployment in metaverse learning environments. By implementing multimodal interaction and adaptive decision-making mechanisms via three agent roles (teaching assistant/learning companion/management agent), we construct an immersive learning ecosystem and establish a context-aware paradigm for intelligent interactive education tailored to Chinese engineering education contexts. This research pioneers an innovative approach to cultivating engineering talent for the AI era through transformative learning methodologies.在高等教育数字化转型的背景下，工程教育面临着理论与实践脱节以及技术工具采用门槛高等关键挑战。本研究开发了一个三元框架（“功能分解 - 过程建模 - 数据治理”）用于教育代理系统，该框架基于 Coze 平台构建，并严格遵循中国新兴工程教育（3E）倡议的标准，旨在元宇宙学习环境中部署。通过实施多模态交互和自适应决策机制，借助三个代理角色（教学助理/学习伙伴/管理代理），我们构建了一个沉浸式学习生态系统，并为中国工程教育环境建立了智能互动教育的情境感知范式。这项研究开创了一种通过变革性学习方法培养人工智能时代工程人才的创新途径。