Video Guides on Head-Mounted Displays: The Effect of Misalignments on Manual Task Performance

Pathirannahalage Shalika Prabhani Pathirathna (1451206)

Head-Mounted Displays (HMDs), such as Google Glass, Oculus Rift, are becoming widely available at affordable prices. Manual task support is a common use of HMDs. There are two fundamental ways of displaying instructions. Either registered to the real environment, following the Augmented Reality (AR) paradigm; or unregistered, following the Wearable Computing paradigm. AR-style instructions are self-aligning and are therefore most effective; however, they require significant effort for authoring and prepraration the environment. On the other hand, unregistered instructions are much easier to provide to users; however, misalignments are unavoidable and reduce task performance.

The goal of our investigations is to quantify how performance deteriorates with increased misalignments. In particular, we investigate the impact of two different misalignments: First, between the HMD's and the video guide's view-plane normals (World space Misalignment: WM); Second, between the HMD's and the video guide's up vectors (Screen space Misalignment: SM). We present the results of two experiments that looked at how task completion times were affected by different misalignments. We used a Lego block placement task according to a video guide where the video guide displayed on the top right corner of the HMD view for both experiments. In the first experiment, we only varied WM, in the second experiment, we additionally varied SM.

Our results show that task completion time is significantly correlated to the absolute difference between WM and SM. This is in line with the underlying mathematical concepts of projective geometry. However, to the best of our knowledge, this has never been verified experimentally. We believe that our results can be applied to a wide range of task support applications. For example, we can now quantify whether a task can be successfully supported with an unregistered video guide; or, whether it requires a more demanding AR-style guide.