A Computational Framework for Multi-Dimensional Virtual Reality System Evaluation: A Case Study in Maritime Safety Training

Abstract

Deploying immersive virtual reality (VR) in safety-critical training is not a simple matter of confirming that the software runs. Standard usability tests overlook the overlap between cognitive load, physiological discomfort, and actual learning success that VR produces at the same time. This study presents a three-tier evaluation framework that merges the System Usability Scale (SUS), the Simulator Sickness Questionnaire (SSQ), and the Learning Experience Questionnaire (LEQ) into a single, repeatable data pipeline. Seventy cadets at a maritime academy completed a twenty-minute Abandoning Ship simulation, after which their responses were collected and processed. The SUS reached a mean of 71.07 (SD = 10.49), placing the software above the established average benchmark of 68 and at the boundary of the "Good" adjective rating [19]. Meanwhile, the SSQ showed a mean of 13.74 (SD = 3.31), suggesting that physical discomfort stayed low, though oculomotor strain emerged as the highest-scoring subscale (M = 1.55). A Pearson correlation between SUS and SSQ scores yielded r = -0.40 (p < 0.001), confirming a statistically significant link between how easy the tool is to use and how much sickness it induces. The LEQ averaged 4.56 out of 5.00 (SD = 0.41), indicating high perceived learning effectiveness. The evaluation is therefore neither a pure software check nor a purely physiological assessment, but a combination of both. Taken together, the results show that the proposed framework can quantify usability, physiological impact, and learning effectiveness for VR training systems, supporting the goal of ensuring that smart-society technology works reliably for the people who use it.