E-Locus - Institutional Repository of the University of Crete

Home Search

Results - Details

Search command : Author="Παπαγιαννάκης" And Author="Γεώργιος"

Current Record: 10 of 28

[Add to Basket]

Identifier

000434161

Title

An analytics algorithm for performance assessment in VR training

Alternative Title

Ένας αλγόριθμος αναλυτικών στοιχείων για αξιολόγηση της απόδοσης εκμάθησης στην εικονική πραγματικότητα

Author

Κέντρος, Μιχαήλ Γ.

Thesis advisor

Παπαγιαννάκης, Γεώργιος

Reviewer

Τζίτζικας, Ιωάννης
Ρούσσος, Αναστάσιος

Abstract

Virtual Reality (VR) hardware and software solutions are rapidly evolving, providing developers with innovative technologies and tools to build applications, in the form of educational virtual experiences. In addition, the popularity of learning and training through such realistic and low-cost VR simulations is growing. Correspondingly, computer graphics and developers utilize the existing VR tools in combination with 3D game engines to create VR immersive experiences tailored around education. However, the current platforms for 3D interactive environments focus on producing mostly embedded gamification tools that serve the entertaining capabilities of VR technology. Task performance assessment is a vital part of the learning process, and by providing valuable feedback it guides the learner towards improvement. In this thesis we addressed this issue, by developing a platform and algorithms that enable developers to accurately, rapidly and systematically author the automatic task performance assessment process of VR training scenarios. We introduce three generalized components, non-dependent to the context of the VR simulation, a) the VR analytics assessment framework, b) a Machine Learning (ML) algorithm capable of VR assessment and c) the VR Session Logger. In more detail, our analytics assessment framework utilizes user analytics for computing the user’s score through an authoring tool for defining performance evaluation rules, whereas by employing supervised ML, our agent is capable of learning these rules directly from a subject matter expert’s (SME) VR data. Furthermore, we present our novel algorithm for logging accurately VR sessions by recording the user’s movement and tracking the resulting effects.

Language

English

Subject

Logging

Machine learning

Task assessment

Virtual reality

Αναλυτικά στοιχεία

Καταγραφή δεδομένων

Μηχανική μάθηση

Issue date

2020-11-27