Mixed Reality Agents for Automated Mentoring Processes.

Mentoring processes can enhance education by providing personalized advice and feedback to students. A challenge of mentoring is that with a rising number of students, more mentors are required. As it is oftentimes infeasible to employ such a high number of mentors, automated tools can support the activities of mentors by e.g., answering common questions. However, such tools can impact the students' engagement as they can feel impersonal. Therefore, we developed mixed reality mentoring agents. They personify these automated tools, can interact directly with the students, and demonstrate practical tasks to them as a guide. On the technical level, this is realized by a behavior tree structure with blackboards that simulate the agent's memory. With such a visual representation of the behavior, developers, teachers, and mentors alike can edit and define the mentoring capabilities of the agent. The implementation results are open-source and we added them to our Virtual Agents Framework that allows developers to quickly add agents to cross-platform mixed reality applications. Moreover, we conducted a user study with the mentoring prototype. The results are promising as students perceived the mixed reality agents in a positive way, with high usability, and as helpful advisors. Therefore, mixed reality mentoring agents have the potential to become widespread companions for students during their studies.