CAREER: Improving the Development Process for Context-Aware Systems with Integrated Capture and Playback

Regents Of The University Of Michigan - Ann Arbor, Ann Arbor MI

Investigators

Abstract

Cheap sensors, high-speed wireless networks, and mobile device technologies are opening up new possibilities for human-computer interaction and are enabling important applications in areas such as health care, collaborative work, and sustainable resource use. Yet we still lack appropriate tools and methods for the design and development of applications that take advantage of these capabilities, hampering our ability to realize the technology's potential. When working with "context-aware" systems -- i.e., systems that sense and respond to the situations in which they are used, such as the user's current location, concurrent activities, or social setting -- it can be especially challenging to evaluate early-stage prototypes due to the difficulty of re-creating the anticipated context of use during development time. As a result, designers are forced to invest excessive effort into building robust, deployable prototypes early in the development process, resulting in premature commitment to inadequately explored design choices and an inability to apply best practices for user-centered design. The goal of this CAREER project is to improve user-centered software development practices for context-aware applications by providing support for the systematic capture and reuse of contextual data throughout the development process. While previous efforts to support context-aware development have sought to make it easier to take prototypes into the field for testing, this approach seeks to "bring the field into the lab" by providing continuously available representations of an application's anticipated context of use. Such representations can be used for exploring and validating design alternatives with as little effort as possible. The PI's RePlay system provides baseline support for the capture and playback of sensor traces representing an application's context, and will be extended through this research to include support for the capture and use of large sensor trace datasets; rapid, parallel prototyping of both interactive and infrastructure components of context-aware systems; and the ability to re-create complex contextual conditions during controlled user tests to an extent not currently possible. Broader impacts: The PI's educational mission is to prepare rising HCI professionals for the constantly changing world of technology they will face throughout their career. As part of this project, he will develop teachable methods for integrating data capture, context representation, and novel forms of user testing into software design and development practice. These methods will be incorporated into existing design and evaluation courses at the graduate and undergraduate levels. Broader impacts will be obtained by sharing course materials, along with the tools described above, with educators and practitioners via the web. The PI also plans to present tutorials and workshops on the use of capture and playback tools during design at meetings and conferences hosted by professional design organizations.