<![CDATA[The rapid digital transformation in higher education has accelerated the adoption of Learning Management Systems (LMS) as a primary platform for online learning, including in the Teacher Professional Education (PPG) program at Universitas Negeri Malang. However, synchronous LMS implementation still suffers from low student engagement, particularly in vocational fields such as automotive education which demand procedural knowledge and hands-on practice. Learning interactions tend to be one-way, lack emotional sensitivity, and reduce student participation. This study develops and evaluates an Emotionally Responsive Pedagogical Agent (PAER) to enhance engagement in synchronous LMS learning. PAER is an AI- and NLP-based virtual assistant designed to detect and respond to learners’ emotional states in real time, providing empathetic interaction and instructional support. The research adopts the ADDIE development model, consisting of needs analysis, emotional interaction design, prototype development, implementation in the PPG Automotive class, and evaluation. PAER effectiveness was measured across affective, cognitive, and behavioral engagement. Results show improved engagement: affective (61.3 to 83.4; N-Gain 0.57), cognitive (63.8 to 86.1; N-Gain 0.61), and behavioral (58.2 to 81.6; N-Gain 0.56), with an overall N-Gain of 0.58 (moderate-high category). These results indicate that emotionally adaptive learning agents can significantly improve interaction quality in LMS-based learning and foster more personalized and human-centered learning. This study contributes to the development of emotional AI in education and offers a scalable model for enhancing online pedagogy in vocational teacher training.]]>
