Virtual avatars for instruction and student response during streaming video-based on-line teaching

This paper describes the design, deployment, and evaluation of an avatar-based system for real-time interaction between the instructor and students in a streaming video-based on-line teaching scenario. The sudden, massive migration to on-line teaching caused by the COVID-19 pandemic caused widespread adoption of streaming video-based on-line teaching methods. However, many subtle features of in-person teaching is lost in a video stream, including physical cues from the instructor or anonymity in student response. In our experience, these changes significantly hinders effective teaching in the on-line environment. We describe our to remedy these issue using a population of virtual avatars overlaid on the video stream to both aid instructor expression, as well as lower the hurdle of student response. Anonymous student polls indicate this is a promising, low-overhead method of improving streaming video-based on-line teaching.