Computational interaction techniques for 3D selection, manipulation and navigation in immersive VR

3D interaction provides a natural interplay for HCI. Many techniques involving diverse sets of hardware and software components have been proposed, which has generated an explosion of Interaction Techniques (ITes), Interactive Tasks (ITas) and input devices, increasing thus the heterogeneity of tools in 3D User Interfaces (3DUIs). Moreover, most of those techniques are based on general formulations that fail in fully exploiting human capabilities for interaction. This is because while 3D interaction enables naturalness, it also produces complexity and limitations when using 3DUIs. In this thesis, we aim to generate approaches that better exploit the high potential human capabilities for interaction by combining human factors, mathematical formalizations and computational methods. Our approach is focussed on the exploration of the close coupling between specific ITes and ITas while addressing common issues of 3D interactions. We specifically focused on the stages of interaction within Basic Interaction Tasks (BITas) i.e., data input, manipulation, navigation and selection. Common limitations of these tasks are: (1) the complexity of mapping generation for input devices, (2) fatigue in mid-air object manipulation, (3) space constraints in VR navigation; and (4) low accuracy in 3D mid-air selection. Along with two chapters of introduction and background, this thesis presents five main works. Chapter 3 focusses on the design of mid-air gesture mappings based on human tacit knowledge. Chapter 4 presents a solution to address user fatigue in mid-air object manipulation. Chapter 5 is focused on addressing space limitations in VR navigation. Chapter 6 describes an analysis and a correction method to address Drift effects involved in scale-adaptive VR navigation; and Chapter 7 presents a hybrid technique 3D/2D that allows for precise selection of virtual objects in highly dense environments (e.g., point clouds). Finally, we conclude discussing how the contributions obtained from this exploration, provide techniques and guidelines to design more natural 3DUIs.