Touchless gestures are used for input when touch is unsuitable or unavailable, such as when interacting with displays that are remote, large, public, or when touch is prohibited for hygienic reasons. Traditionally user input is spatially or semantically mapped to system output, however, in the context of touchless gestures these interaction principles suffer from several disadvantages including memorability, fatigue, and ill-defined mappings. This thesis investigates motion correlation as the third interaction principle for touchless gestures, which maps user input to system output based on spatiotemporal matching of reproducible motion. We demonstrate the versatility of motion correlation by using movement as the primary sensing principle, relaxing the restrictions on how a user provides input. Using TraceMatch, a novel computer vision-based system, we show how users can provide effective input through investigation of input performance with different parts of the body, and how users can switch modes of input spontaneously in realistic application scenarios. Secondly, spontaneous spatial coupling shows how motion correlation can bootstrap spatial input, allowing any body movement, or movement of tangible objects, to be appropriated for ad hoc touchless pointing on a per interaction basis. We operationalise the concept in MatchPoint, and demonstrate the unique capabilities through an exploration of the design space with application examples. Finally, we explore how users synchronise with moving targets in the context of motion correlation, revealing how simple harmonic motion leads to better synchronisation. Using the insights gained we explore the robustness of algorithms used for motion correlation, showing how it is possible to successfully detect a user's intent to interact whilst suppressing accidental activations from common spatial and semantic gestures. Finally, we look across our work to distil guidelines for interface design, and further considerations of how motion correlation can be used, both in general and for touchless gestures.
|Publication status||Published - 2020|