A comparison of interaction models in Virtual Reality using the HTC Vive

Sammanfattning: Virtual Reality (VR) is a field within the gaming industry which has gained much popularity during the last few years. This is caused mainly by the release of the VR-headsets Oculus Rift [1] and HTC Vive [2] two years ago. As the field has grown from almost nothing in a short time there has not yet been much research done in all VR-related areas. One such area is performance comparisons of different interaction models independent of VR-hardware. This study compares the effectiveness of four software-based interaction models for a specific simple pick-and-place task. Two of the interaction models depend on the user moving a motion controller to touch a virtual object, one automatically picks them up on touch, the other requires a button press. The other two interaction models have the user move a laser pointer to point at an object to pick it up. The first has the laser pointer emitted from a motion controller and the second has it emitted from the user’s head. All four interaction models use the same hardware, the default HTC Vive equipment. The effectiveness is measured in three metrics, time to complete the task, number of errors made during the task, and the amount of participant enjoyment rated on a scale from one to five. The first two metrics are measured through an observational experiment where the application running the virtual environment also logs all relevant information. The user enjoyment is gathered through a questionnaire the participant answers during the experiment. These are the research questions: • How do the interaction models compare in terms of accuracy and time efficiency when completing basic pick and place tasks in this experiment? • Which interaction models are subjectively more enjoyable to use according to participants? The results of the experiment are displayed as charts in the results chapter and then further analysed in the analysis and discussion chapter. Possible sources of error and theories about why the results turned out the way they did are also discussed. The study concludes that the laser pointer based interaction models, 3 and 4, were much less accurate than the handheld interaction models, 1 and 2, in this experiment. All interaction models except 4 achieved about the same test duration while interaction model 4 lagged several seconds behind. The participants liked interaction model 1 the most, followed closely by 3. They disliked 4 the most and rated 2 at a point in the middle of the rest.