Probabilistic User Interface Design for Virtual and Augmented Reality Applications

The central hypothesis of this thesis is that probabilistic user interface design provides an effective methodology for delivering productive and enjoyable applications in virtual reality (VR) and augmented reality (AR). This investigation is timely given the recent emergence of mass-market virtual and augmented reality head-mounted displays and growing demand for tailored applications and content. The design guidance for building compelling and productive applications for these environments is, however, currently lagging the pace at which the underlying technology is maturing. This is problematic given important differences between designing conventional 2D interfaces and interactions and their embodied 3D counterparts. This dissertation investigates probabilistic user interface design as a method for solving many of the novel challenges encountered when developing applications for VR and AR. Probabilistic user interface design seeks to model the uncertain events in a system and identify, implement and validate strategies that drive improved system performance. This thesis addresses four research questions by applying a probabilistic treatment in four distinct but closely related case studies. These four case studies are selected to illustrate the flexibility and unique benefits offered by this method. Research Question 1 asks how the probabilistic qualities of an interface can be determined and how this can inform design. This question is investigated in the context of text entry in VR with a probabilistic characterisation performed on two fundamental design choices. Research Question 2 relates to the challenge of adapting AR applications to deployment contexts not knowable at design time. A study in which crowdworkers are employed to build a probabilistic understanding of the requirements for contextually adaptive AR answers this question. The text entry theme is revisited in answering Research Questions 3 which asks how high levels of input noise can be mitigated through inference. A probabilistic text entry method specifically tailored for use in AR is implemented and evaluated. Finally, Research Question 4 asks how the high dimensional design space in AR and VR applications can be efficiently explored to support ideal design choices. Interface refinement through probabilistic optimisation and crowdsourcing is shown to be highly efficient and effective for this purpose. A probabilistic treatment in the design process has many potential benefits, principle among which is increased robustness to circumstances unanticipated at design time. This thesis contributes to the toolset and guidance available to designers and supports the development of next generation user interfaces specifically tailored to virtual and augmented reality.

Name dc:type.qualificationname

Doctor of Philosophy (PhD)

Level dc:type.qualificationlevel

Doctoral

Grantor dc:publisher.institution

University of Cambridge

Year dc:date.issued

2020

Author dc:creator

• Dudley, John James

Advisor dc:contributor.advisor

• Per Ola, Kristensson

• Virtual Reality
• Augmented Reality
• Mixed Reality
• Probabilistic User Interface Design
• Text Entry

Licence

• https://www.repository.cam.ac.uk/bitstreams/6cfa1df4-d593-4b73-bb4a-baaa95a9681b/download
• https://www.rioxx.net/licenses/all-rights-reserved/

Language dc:language

en

Author Identifier

0000-0001-6692-4853

OAI identifier oai:identifier

oai:www.repository.cam.ac.uk:1810/304577