Toward Disambiguating Multiple Selections for Frustum-Based Pointing

These environments often simulate or augment the real world, and a part of that simulation is the ability to select objects for observation and manipulation. Many user interfaces for these applications depend on six-degree-of-freedom tracking devices. Such devices have limited accuracy and are susceptible to noise, giving an imprecision that makes object selections difficult and hard to repeat. This difficulty is amplified when the user’s viewpoint is also tracked, meaning the user must compensate for noise from both the head tracker and the pointing device when performing object selection. Also, users may experience fatigue when using handheld pointing devices for extended periods, creating error even if the tracking technology were perfect. This paper presents a pointing-based probabilistic selection algorithm that addresses some of the ambiguities associated with tracking and user imprecision. It performs multiple selections by considering a frustum along the user’s pointing direction and the hierarchical structure of the database. It assigns probabilities that the user has selected particular objects using a set of low-level 3D intersection-based selection techniques and the relationship of the objects in a hierarchical database, and makes the final selection using one of several weighting schemes. We performed several experiments to evaluate the low-level selection techniques, tested several weighting schemes for the integration algorithm, and we show that the algorithm is effective at disambiguating multiple selections.