Context-Aware Markerless Augmented Reality for Shared Educational Spaces

Abstract

In order for markerless augmented reality (AR) to reach its full potential in educational settings it must be able to adapt to the wide range of possible environments, devices and user cognitive states that affect learning outcomes. Shared educational spaces, such as classrooms, art galleries, museums, teaching hospitals and wildlife centers present enticing opportunities in terms of specialized AR applications, existing infrastructure to leverage, and large numbers of AR sessions to gather data on. In this work we make feasible the challenging concept of context-aware AR through the use of a ‘local expert’, which learns the optimal configuration of AR algorithms and virtual content for the specific educational space and use case for which it is implemented. To compute insights from multiple AR devices, enable timely responses to fast-changing user cognitive states, and ensure the security of sensitive user data, we propose an edge-computing architecture, in which storage and computation related to our local expert is performed on a server on the same local area network as the mobile AR devices.