Animating Avatars in Education Using Machine Learning

For new and existing educators, using virtual characters as pedagogical agents has been shown to increase learner engagement; however, many perceive this technique as costly and complex. More recent advances in machine learning-based facial action coding recognition have made implementing virtual characters into content more accessible and affordable than ever. Introduction to Virtual Avatars for Instruction is a course designed to bridge this gap and change educators’ attitudes about virtual character-led instruction. The course aimed to provide learners with the key technologies, terms, and examples needed to add virtual characters to instructional content. Key materials included didactic short-form virtual videos of pedagogical agent-led instruction, and quizzes, Discord-based peer reflections, surveys, and student demonstrations were utilized to evaluate the course’s usability and learning effectiveness. Results from participants (n = 15) showed an increased understanding of key terms and concepts; however, participants faced challenges when asked to complete peer reflection and commentary. With the completion of peer activities, students successfully demonstrated their knowledge and use of core technologies. Attitudinal changes were also observed, showing a shift away from concerns regarding the cost and skills required to implement virtual characters. With this new outlook, course materials could be expanded to include content about authoring techniques and character creation, giving educators tools to create engaging pedagogical agents and instructional content more readily.

Course link:https://ltec.fun/courses/introduction-to-virtual-avatars-for-instruction/