Three-dimensional interaction with virtual objects is one of the aspects that needs to be addressed in order to increase the usability and usefulness of virtual reality. Human beings have difficulties understanding 3D spatial relationships and manipulating 3D user interfaces, which require the control of multiple degrees of freedom simultaneously. Conventional interaction paradigms known from the desktop computer, such as the use of interaction devices as the mouse and keyboard, may be insufficient or even inappropriate for 3D spatial interaction tasks. The aim of the research in this thesis is to develop the technology required to improve 3D user interaction. This can be accomplished by allowing interaction devices to be constructed such that their use is apparent from their structure, and by enabling efficient development of new input devices for 3D interaction. The driving vision in this thesis is that for effective and natural direct 3D interaction the structure of an interaction device should be specifically tuned to the interaction task. Two aspects play an important role in this vision. First, interaction devices should be structured such that interaction techniques are as direct and transparent as possible. Interaction techniques define the mapping between interaction task parameters and the degrees of freedom of interaction devices. Second, the underlying technology should enable developers to rapidly construct and evaluate new interaction devices. The thesis is organized as follows. In Chapter 2, a review of the optical tracking field is given. The tracking pipeline is discussed, existing methods are reviewed, and improvement opportunities are identified. In Chapters 3 and 4 the focus is on the development of optical tracking techniques of rigid objects. The goal of the tracking method presented in Chapter 3 is to reduce the occlusion problem. The method exploits projection invariant properties of line pencil markers, and the fact that line features only need to be partially visible. In Chapter 4, the aim is to develop a tracking system that supports devices of arbitrary shapes, and allows for rapid development of new interaction devices. The method is based on subgraph isomorphism to identify point clouds. To support the development of new devices in the virtual environment an automatic model estimation method is used. Chapter 5 provides an analysis of three optical tracking systems based on different principles. The first system is based on an optimization procedure that matches the 3D device model points to the 2D data points that are detected in the camera images. The other systems are the tracking methods as discussed in Chapters 3 and 4. In Chapter 6 an analysis of various filtering and prediction methods is given. These techniques can be used to make the tracking system more robust against noise, and to reduce the latency problem. Chapter 7 focusses on optical tracking of composite input devices, i.e., input devices 197 198 Summary that consist of multiple rigid parts that can have combinations of rotational and translational degrees of freedom with respect to each other. Techniques are developed to automatically generate a 3D model of a segmented input device from motion data, and to use this model to track the device. In Chapter 8, the presented techniques are combined to create a configurable input device, which supports direct and natural co-located interaction. In this chapter, the goal of the thesis is realized. The device can be configured such that its structure reflects the parameters of the interaction task. In Chapter 9, the configurable interaction device is used to study the influence of spatial device structure with respect to the interaction task at hand. The driving vision of this thesis, that the spatial structure of an interaction device should match that of the task, is analyzed and evaluated by performing a user study. The concepts and techniques developed in this thesis allow researchers to rapidly construct and apply new interaction devices for 3D interaction in virtual environments. Devices can be constructed such that their spatial structure reflects the 3D parameters of the interaction task at hand. The interaction technique then becomes a transparent one-to-one mapping that directly mediates the functions of the device to the task. The developed configurable interaction devices can be used to construct intuitive spatial interfaces, and allow researchers to rapidly evaluate new device configurations and to efficiently perform studies on the relation between the spatial structure of devices and the interaction task.

, , ,
R. van Liere (Robert) , J.J. van Wijk (Jack)
Technische Universiteit Eindhoven
Visualization and 3D User Interfaces

van Rhijn, A. (2007, January 18). Configurable Input Devices for 3D Interaction using Optical Tracking. Configurable Input Devices for 3D Interaction using Optical Tracking.