ProxemicUI: Iterative Design and Evaluation of a Flexible and Generic Framework for Proxemics Aware Applications

Abstract

Systems that respond to spatial configurations of people, devices, and objects (sometimes called proxemics-aware systems) have long been explored in HCI research and are now finding commercial applications. While state-of-the-art tracking systems and toolkits have been demonstrated the ability to provide proxemics data, there is still a considerable gap between what existing systems support and what the developers require to define their own custom proxemics events (e.g., to indicate that complex spatial configurations are meaningful) rather than conduct manual tests in the UI layer after low-level events are triggered. In addition, proximity-related events are often intertwined with other events, such as interaction with application interfaces and notifications from smart appliances. While machine learning techniques might detect complex spatial configurations, more research is needed to determine what kinds of input data are required to train an accurate and adaptable classifier. In response to these issues, I developed ProxemicUI, an object-oriented framework that provides a simple but powerful rule-based format for expressing proxemics relationships, an event-driven model for responding to their occurrence, a mechanism for integrating non-proxemics data and events, and extensibility through clearly articulated extension points. I evaluate ProxemicUI through a set of proof-of-concept applications, a comparative code review study with 18 programmers, and the integration of ProxemicUI into an augmented reality storytelling tool, working with three authors and three developers. The results demonstrate that ProxemicUI supports rapid prototyping of systems that can respond to complex proxemics events across various contexts.