An adaptive and customizable feedback system for VR-based training simulators

This paper describes a proposal to build an intelligent feedback selection system for Virtual Reality-based training simulators. The system is aimed at generating multimodal feedback in realtime for advising the students while training with the simulator. Focused on driving tasks, we analyze how to customize the system to exhibit different behaviors. We examine educational and human factors that have influence on the behavior, so that the instructors can use or refine the behavior they prefer in each training session. The selection process is based on the analysis of the information coming from a diagnostic component and adapts the feedback to the performance of each student, since the process takes into account whether previous feedbacks were ineffective. The objective is to emulate the behavior of the instructor. In this way, the feedback system can be helpful for him/her, as while the system decides which the appropriate feedback is, the instructor can focus on other instructional tasks.