An intelligent tutoring system for a virtual reality procedural training project

Abstract

The key topic of this master work lies at the intersection of 2 different areas: intelligent tutoring systems and virtual reality. As such, this document will cover both areas and dive into some of the problems faced when implementing a software system at their intersection.
A real-life project, SIEMA, is the source of context, business needs and constraints for this master work. SIEMA was initiated by a car seat manufacturing company (further in the text - SeatInc), that among other things wants to improve following aspects of training program:
• Improve student knowledge of the domain vocabulary and assembly process steps.
• Increase number of students that can go through training program simultaneously.
• Increase the effective time that each student can spend practicing the assembly process.
• Reduce the number of new candidates that are rejected after attempting practical exercises.
• Reduce the time and effort required to re-train existing employees using new processes.
This master thesis will use above objectives as some of the drivers in order to demonstrate how automated procedural training can be applied to replace or assist human trainers in order to improve training efficiency.
Most of the work done in this master from the domain of intelligent tutoring systems was based on the results of PhD thesis[1] of another UPM alumni – Diego Riofrío. Diego Riofrío proposes a tutoring strategy (Collective Student Model) that applies educational data mining (EDM) to predict student performance and adjust tutoring based on the predictions. In addition, thesis outlines a strategy for combining multiple tutoring strategies in a single tutoring component using Tutoring Coordinator. This document will explain core ideas and concepts behind Collective Student Model and Tutoring Coordinator and will pay special attention to implementation details such as design and problems faced and how they were solved.
This master work pursues following goals:
• Propose a viable system architecture for SIEMA project based on the goals of the project
• Implement tutoring strategy based on the collective student model
• Produce a viable tutoring coordinator implementation by combining tutoring strategy based on collective student model with existing reactive tutor.
Next, section 2 of this document addresses state of the art in the relevant fields while section 3 will focus on the SIEMA project and set the context in which further work will be performed. Section 4 of this document will present software engineering practices used during implementation of this project. Section 5 will address the area of Intelligent Tutoring Systems, introduce pre-exsiting components use for development, outline issues faced during the implementation and present solutions and extensions that were implemented in this master work in depth. Finally, section 6 will present conclusions and future work.