Immersieve evaluaties in virtuele realiteit

In the user-centred design process, iteratively building prototypes and conduct usability tests is a critical step. The use of VR to evaluate the usability of these prototypes is gaining increased research attention. We refer to this type of usability studies carried out in VR as "Immersive Virtual Reality Evaluations" (IVREs). The virtual model used in IVRE to represent the physical prototype that is the object of the evaluation, is termed a "virtual twin", closely replicating its physical counterpart in terms of appearance, interactive capabilities, and usability. Previous research has shown that IVREs have the potential to yield results transferable to the real world, stage field studies in VR that are difficult to replicate in real life, identify potential problems at an early stage before building a physical prototype, and simulate the interaction and appearance of physical devices at a lower cost than creating physical prototypes. However, open research questions remain in the field of IVREs. This PhD project focuses on the following three research questions: RQ1: Conventional usability evaluation methods in IVREs; Can we use conventional usability evaluation method in VR and obtain results comparable to performing the evaluation on a physical prototype? RQ2: Virtual Twin Fidelity in IVREs; Does the fidelity of the VR implementation of a prototype affect the number and severity of issues found in a usability evaluation? RQ3: Quantifying Virtual Twin Similarity; Can we quantify the perceptual similarity of a virtual twin to its physical counterpart in a simple virtual environment? To address these research questions, we designed and conducted eight user studies, collecting and analysing both quantitative and qualitative data. To answer the first research question, we examined the performance of Heuristic Evaluation (HE), Think Aloud Protocol (TAP), and standard usability questionnaires in IVREs. We found no significant differences in terms of severity and number of identified usability problems with HE and TAP, nor in the scores from standard questionnaires. Additionally, we demonstrated the integration of IVRE into the development process of an AR product. To explore the second research question, we designed a user study to investigate the impact of different graphic and interaction fidelities of the virtual twin on evaluation results in IVRE. While differences in graphic fidelity (colour-based and physically based rendering) were difficult to identify, participants' comprehension of operations could be misled by supernatural interactions compared to natural ones. This suggests that natural interaction techniques, which closely approximate realworld product interactions, should be implemented. For the third research question, we developed a questionnaire with 21 items across four categories to quantify the similarity of the virtual twin. This work also summarises the lessons learned and opportunities discovered while conducting IVREs for product designers, researchers, and other stakeholders.