Topiary: A Tool for Prototyping LocationEnhanced Applications

1
Topiary A Tool for Prototyping Location-Enhanced
Applications
UC Berkeley UC Berkeley Intel / UW
Yang Li Jason Hong James Landay
2
Location-Enhanced Applications

• Can provide useful services by leveraging
  knowledge about the location of people, places,
  and things
• Require a high level of technical expertise to
  build
• Hard to prototype, evaluate, and iterate on
  designs
• Cannot be tested with end-users until actually
  deployed
• Goal of Topiary
• Allow designers to quickly prototype test
  location-enhanced applications
• no application or infrastructure development
  required
• enable designers to get early feedback from users

3
Iterative Interaction Design
Design
Prototype
Evaluate
4
Examples of Prototyping Tools

• SILK
• Prototyping Graphical User Interfaces
• DENIM
• Informal Web Site Design Over 46000 downloads
  since 2002

5
Research Approaches for Designing
Location-Enhanced Prototyping Tools

• Traditional approach
• Build the tool based on the observation of
  existing practices
• Not possible few experienced designers in this
  emerging area
• Our approach
• Analyze feature set location status, finders,
  active maps, triggers, way finding, resource
  allocation, tagging
• Combine design practices of other domains (e.g.,
  GUIs SUIs)
• Conduct Iterative paper prototyping on the tool
  design evaluate prototypes with both UI
  designers Ubicomp researchers

6
Research Challenges
Enable designers to easily test and analyze a
design in realistic situations
Enable designers to efficiently explore a much
wider input space
Modeling Location Contexts
Testing and Analyzing a Design
Specifying Location-Enhanced Behaviors
Enable designers to specify interaction sequences
that integrate both implicit and explicit
interactions
7
Our Solution Topiary
Enable designers to easily test and analyze a
design in realistic situations
Enable designers to efficiently explore a much
wider input space
Modeling Location Contexts
Testing and Analyzing a Design
Specifying Location-Enhanced Behaviors
Enable designers to specify interaction sequences
that integrate both implicit and explicit input
8
ActiveMap WorkspaceModeling Location contexts

• Model a geographical area and create entities
• Demonstrate location contexts
• Capture Scenarios

9
ActiveMap WorkspaceModeling Location contexts
10
ActiveMap WorkspaceModeling Location contexts
Indicates context also uses temporal information

• ActiveMap workspace gives designers a visual
  language for specifying these contexts

11
ActiveMap WorkspaceModeling Location contexts
12
Scenarios

• Represent a collection of location contexts
• Used as triggers and conditions for specifying
  location-enhanced interactions

13
Capturing Scenarios
14
Abstracting Scenarios
15
Storyboard WorkspaceLocation-Enhanced
Interactions
Explicit Input
Actions
16
Storyboard WorkspaceSpecifying Location-Enhanced
Behaviors

• Pages
• Explicit links and implicit links
• Built-in context components
• Catching conflicts in design

17
Storyboard WorkspaceCreating Pages and Links
18
Storyboard WorkspaceExplicit Links
19
Storyboard WorkspaceImplicit Link
20
Storyboard WorkspaceBuilt-in Context Components

• Display spatial and temporal information
• Reuse typical location-enhanced interactions
• Enable continuous interactions

21
Built-in Context ComponentsNearest Entities
Component


22
Built-in Context ComponentsActiveMap Component


23
Test WorkspaceTesting and Analyzing a Design
End-user UI
Wizard UI
24
Test WorkspaceTesting and Analyzing a Design
25
Detecting Conflicts in DesignExclusion in a Link
26
Detecting Conflicts in DesignExclusion Detection
Constraint Satisfaction Problem (CSP)
Exclusion Detection
Geometric CSP
Geometric CSP
Bob is far from the Kitchen
Knowledge of spatial relationships
Detecting Exclusion
Boolean CSP
Boolean CSP
Its still NP-complete, but there are algorithms
for this.
27
Detecting Conflicts in DesignAmbiguity between
Links
28
Implementation

• Built in Java 2 SDK v1.4.2 with 343 Java classes
  and approximately 23000 physical lines of source
  code
• Using SATIN, a toolkit for sketch-based user
  interfaces (Ex. Pie menu and navigation gestures)
• End-User UI a thin client with 11 classes
  compatible with JRE1.1
• Communication between Wizard UI and End-User UI
  Java object serialization and network socket
• Tested with wireless connection over IEEE802.11
  (access point and peer-to-peer)

29
Evaluation

• Settings
• Tasks
• Participants
• 2 ubicomp researchers, 3 professional UI
  designers (one from industry) and 2 undergraduate
  students who took a UI design course
• Results
• Informal user study on Topiary gave positive
  feedback

30
Conclusion

• Topiary is the first tool for prototyping
  location-enhanced applications
• Demonstrating location contexts via an Active Map
• Specifying location-enhanced behaviors via
  enhanced storyboards
• Testing and analyzing a design by Wizard of Oz

31
Future Work

• Specifying the ambiguity of sensor data
• Enabling more kinds of contextual information
• Improving storyboard scalability
• Incorporating location infrastructures
• Place Lab larger scale and more realistic
  testing

32

• Thanks!
• http//guir.berkeley.edu/projects/topiary