PhD Oral Exam - Mohammad Raihanul Bashar, Computer Science

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Extended Reality enables users to enter spatial environments where digital objects can be seen, arranged, and acted upon as if they were part of the surrounding world. However, this promise weakens when users try to select objects inside XR. Targets may appear at different depths, the system may lose track of the user's hands, and tasks may require several objects to be selected as a coherent set rather than just one. Selection is therefore not only a matter of choosing an input technique; it is shaped by the constraints under which perception, tracking, and task structure meet. This dissertation develops a constraint-aware account of 3D selection in XR First, we examine visual depth as a perceptual constraint. Through controller- and gaze-based selection studies, we show that depth affects selection performance systematically and that optical focus demand, expressed in diopters, models this effect better than linear distance alone. Second, we examine hand-tracking failure as a tracking constraint on bare-hand interaction. We present an early warning system that uses headset tracking-confidence data to signal impending tracking loss, improving perceived usability, reducing frustration, and mitigating the task-performance cost of tracking failure. Third, we examine multi-object selection as a task-structure constraint. Through serial and parallel multi-selection studies, we characterize how mode-switching, subselection, boundary representation, input modality, target geometry, and target count shape performance and preference. Together, this thesis shows that effective XRselection is not achieved by a universally best technique. Instead, selection techniques must be designed around the perceptual, tracking, and task-structure constraints that govern interaction in XR.