Spatial input device structure and bimanual object manipulation in virtual environments

Complex 3D interaction tasks require the manipulation of a large number of input parameters. Spatial input devices can be constructed such that their structure reflects the task at hand. As such, somatosensory cues that a user receives during device manipulation, as well as a users expectations, are consistent with visual cues from the virtual environment. Intuitively, such a match between the device's spatial structure and the task at hand would seem to allow for more natural and direct interaction. However, the exact effects on aspects like task performance, intuitiveness, and user comfort, are yet unknown.The goal of this work is to study the effects of input device structure for complex interaction tasks on user performance. Two factors are investigated: the relation between the frame of reference of a user's actions and the frame of reference of the virtual object being manipulated, and the relation between the type of motion a user performs with the input device and the type of motion of the virtual object.These factors are addressed by performing a user study using different input device structures. Subjects are asked to perform a task that entails translating a virtual object over an axis, where the structure of the input device reflects this task to different degrees. First, the action subjects need to perform to translate the object is either a translation or a rotation. Second, the action is performed in the same frame of reference of the virtual object, or in a fixed, separately located, frame of reference.Results show that task completion times are lowest when the input device allows a user to make the same type of motion in the same coordinate system as the virtual object. In case either factor does not match, task completion times increase significantly. Therefore, it may be advantageous to structure an input device such that the relation between its frame of reference and the type of action matches the corresponding frame of reference and motion type of the virtual object being manipulated.