Online training evaluation in VR simulators using Gaussian Mixture Models

Area of Science:

• Simulation and Training Technologies
• Virtual Reality Applications
• Machine Learning in Education

Background:

• Effective evaluation of simulator training is crucial for skill acquisition.
• Virtual reality (VR) offers immersive training environments but lacks standardized evaluation metrics.
• Traditional assessment methods may not capture the nuances of VR-based training performance.

Purpose of the Study:

• To propose a novel, automated approach for evaluating training effectiveness in virtual reality simulators.
• To leverage machine learning for objective performance classification during simulation-based training.
• To establish a robust method for categorizing simulation performance within predefined training classes.

Main Methods:

• Development of a new evaluation framework for virtual reality simulator training.
• Application of Gaussian Mixture Models (GMM) for data modeling and analysis.
• Classification of simulation data into distinct, pre-defined training categories.

Main Results:

• The proposed approach successfully models and classifies simulation data using GMM.
• Demonstrated feasibility of using GMM for objective evaluation of VR training.
• Pre-defined training classes were effectively identified through the GMM classification.

Conclusions:

• Gaussian Mixture Models provide a powerful tool for evaluating virtual reality simulator training.
• This GMM-based approach offers an objective and scalable method for performance assessment.
• The proposed technique enhances the quality and standardization of simulator training evaluation.