Three-dimensional (3D) real-time interaction in virtual environments (VEs) has been a topic of research for a few decades. Interaction methods have evolved in parallel with processing power and motion-tracking technology.
This paper starts with an ample review of the state of the art in 3D interaction design and provides a solution for 3D interaction based on computer vision motion tracking. The authors focus on three common VE tasks: object selection, object manipulation, and navigation.
The experimental design presents a comparison between a 3D glove-based system and the computer vision system based on Microsoft Kinect. Several tasks have been proposed, and the execution time, manipulation time, and the error of accuracy are compared.
A comprehensive set of experiments involving 20 participants shows insignificant differences between the two systems in terms of tiredness, effectiveness, precision, stability, and ease of use. However, comfort, reliability, and handiness are significantly better with the 3D glove system, whereas freedom of movement is improved with the proposed vision-based system.
The results are in line with what one would expect, considering that the vision-based system is tetherless. Additionally, the 3D glove system, although more complex and expensive, does bring the added benefits of effectiveness, precision, and stability.
Fundamentally, the conclusion is that the choice of 3D real-time interaction is task dependent; however, vision-based techniques are more often employed and have better potential in terms of system hardware simplicity, cost, and freedom of movement.