TL;DR
This paper introduces Topiary, a novel prototyping tool designed to simplify the creation of location-enhanced applications. By abstracting away low-level sensor complexities and providing intuitive visual tools for modeling spatial contexts and interactions, Topiary empowers designers to rapidly iterate on location-aware application designs.
What Problem Does This Paper Address?
Location-enhanced applications, which leverage a user's current location and the locations of other entities to augment or streamline interactions, are poised for significant growth. However, building these applications currently demands a high level of technical expertise, particularly in dealing with diverse sensing technologies (like GPS and beacons) and complex spatial reasoning. This technical barrier makes it difficult for designers to quickly prototype, evaluate, and iterate on their designs, often leading to costly changes late in the development cycle. The paper aims to bridge this gap by providing a tool that allows designers to focus on the user experience and interaction design of location-enhanced applications without getting bogged down in low-level implementation details.
Methods
Topiary offers a multi-workspace approach to prototyping:
- Active Map Workspace: Designers create a visual model of the application's environment, defining people, places, and things. This workspace allows for sketching maps or importing existing images as backgrounds. It also supports drawing paths to define road networks for wayfinding features.
- Scenario Producer: This tool enables designers to capture location contexts through a "programming by demonstration" paradigm. By defining spatial relationships (e.g., "Alice is in the Gym," "Bob enters Room 525") and transitions, designers can create reusable scenarios. Proximity regions can also be defined visually.
- Storyboard Workspace: This is where the user interface mockups are created. Designers can sketch pages and define links between them. Crucially, scenarios captured in the Active Map workspace can be used as conditions or triggers for these links, enabling conditional navigation and interaction flows. Topiary supports both explicit links (user-initiated actions) and implicit links (automatically triggered by scenario changes).
- Test Workspace: This workspace allows designers to test their prototypes using a "Wizard of Oz" approach. A "Wizard UI" lets designers simulate location contexts by moving entities on the Active Map, while an "End-User UI" (run on a separate device, like a PDA) displays the application's interface. This allows for real-time testing and observation of how users interact with location-aware features. Topiary can also optionally integrate real sensor data for more realistic testing.
Datasets
Not discussed in the paper.
Key Insights and Ideas
- Abstraction of Sensing Complexity: The core insight is that for early-stage design and prototyping, designers don't need to grapple with the intricacies of actual sensor hardware. Topiary's Wizard of Oz approach effectively fakes sensor data, allowing designers to focus on the implications of location rather than the mechanics of sensing.
- Visual Language for Spatial Context: Topiary introduces a visual language for defining location contexts and scenarios, making it more intuitive for designers to express spatial relationships and dynamic changes in the environment. This moves beyond purely textual or logical representations.
- Scenarios as First-Class Citizens: Scenarios are not just annotations but are integral to the prototyping process, acting as conditions for UI transitions and behaviors. This allows for the creation of truly context-aware interactions.
- Implicit vs. Explicit Links: The distinction between explicit (user-driven) and implicit (context-driven) links in the storyboard is a powerful way to model the unique interaction patterns of location-enhanced applications.
Key Results
An informal evaluation with 7 participants (researchers and interface designers) showed positive reactions to Topiary's concept. Participants generally found the tool understandable and useful, with an average rating of 5.7/7 for understandability, 5/7 for ease-of-use, and 5.9/7 for usefulness. While participants could accomplish most tasks, they noted challenges with managing complex link interactions and ensuring all scenarios were covered in the Storyboard workspace. The authors acknowledge these issues and mention subsequent improvements like the storyboard analysis view and test recording/replay features.
Limitations and Future Work
- Storyboard Complexity: Participants found it challenging to manage a large number of crossing links and ensure all scenarios were covered, especially as the storyboard grew. The authors acknowledge this and suggest future work on hierarchical storyboarding (e.g., using StateCharts) and better awareness of covered scenarios.
- Sensor Ambiguity: Topiary currently doesn't explicitly handle sensor ambiguity, though it indirectly supports it through its sketching model and optional real sensor integration. Future work includes adding basic support features for sensor ambiguity, potentially through error models.
- Limited Context Types: Currently, Topiary primarily supports spatial and temporal contexts. Future work aims to incorporate other contextual information, such as user activity.
- Resource Allocation and Tagging: The tool currently does not explicitly support prototyping features related to resource allocation or tagging, as these are considered more internal computation than direct user interaction.
- Non-Display Output: Topiary is primarily focused on display-based visual output and does not provide explicit hooks for non-display outputs (e.g., turning on lights).
Summary updated