Automating the Design of Visual Instructions

[Summary]  [Publications]  [Other References]  [Documents]  [Schedule]  [People]  [Contact] 

Summary

Visual instructions are a common part of our daily lives. Maps, training manuals, textbooks, architectural plans, scientific papers, and street signs all use visual diagrams to communicate instructions. Yet, even the simplest visualizations typically take hours or days to design by hand, and therefore it is not currently possible to adapt and personalize instructions to the task, person, and situation for which they are eventually used. In contrast, while current computer-generated visualizations can be generated very quickly, these systems disregard many of the cognitive design principles that guide human designers. As a result current computer-generated visualizations can be very difficult to use.

What are needed are fully automated design systems that are capable of producing visual instructions that are as effective as hand-designed instructions. We have recently developed such an automated design system for rendering route maps, that we call LineDrive. LineDrive was originally developed, when we observed that hand-designed route maps (even when quickly sketched) are usally much easier to use than standard computer-generated route maps (e.g. the kind of maps produced as driving directions by MapQuest or MapBlast!). We studied a variety of hand-designed route maps and analyzed cognitive psychology research on how route maps are used to understand why these maps are so effective and enumerate the set of cogntive design principles used in the maps. We then encoded these design principles algorithmically within the LineDrive system.

We believe that the approach we developed for automating the design of route maps is general and can be applied to other visualization domains. For example we might apply this approach to automatically designing other types of cartographic maps, mechanical assembly instructions, logisitics plans and timelines, or architectural plans. Intially we plan to explore how the approach might be applied to the following problems:

  • Automating the design of route maps that adapt to the task, person, or situation.
  • Automating the design of other types of maps: Point location maps, Routes through 3D environments and Dynamic Route Maps.
  • Automating the design of assembly instructions for mechanical objects such as modular furniture, lego, etc.

Publications

Other References

Documents

  • Proposal to Augmented Cognition program (17 MB Microsoft Word format).
  • Maneesh's gcafe presentation from 9-27-01 (34 MB Microsoft Powerpoint format).

Schedule

People

Maneesh Agrawala
Pat Hanrahan
Julie Heiser
Paul Lee
David Salesin
Chris Stolte
Maureen Stone
Diane Tang
Barbara Tversky

Contact

For more information, please contact Maneesh Agrawala (maneesh@graphics.stanford.edu).
Fig 1: LineDrive map and standard MapBlast! map for the same route. Since all the turning points are visible in the LineDrive map it is much easier to use.

Fig 2: Routes through a 3D environment.

Fig 3: Assembly instructions for Lego firetruck.

maneesh@graphics.stanford.edu