Battlespace Visualization
=========================

Real-time, multi-source data.

Focus and tailor visualizations -- users have different tasks, missions, roles.

User is sleep deprived, under high stress, short job cycle.

HCI must be very robust: reduce error rate to zero.

Jove example
------------

?

A great deal is at stake.

User community is diverse in every way.

Mix of 3-d models and iconic representations of units.

There are existing legacy visual languages in some situations
(e.g. operations) but not others (e.g. info warfare).

There is a dramatic need for infovis to discover structure in
apparently unstructured collections of intelligence data.

Integrated Command Environment
------------------------------

Trish Hamburger

www.manningaffordability.com

Visualization capabilities must support: decision-making for the warfighter;
analysis for the intelligence; process for designers.
Must acknowledge and work with human limitations.

Visualization for training debriefs: automated performance measurement,
team performance feedback.

Involve "warfighters" in design process.

Military propaganda...

Five Challenges
---------------

Jeff Grossman, Naval Warfare... (Kathy McDonald)

1. Consistent situation representation.  Information arriving in different
modes (from humans, electronics, imagery, satellites, sensors, intelligence,
local combat direction systems, data links, etc.).  How to register all the
information into a consistent framework?  Achieving the "common picture".
With so many data sources, they may not be consistent.  But what you're
seeing must be consistent with what others are seeing at the same time.
Fusion and synthesis.  Level 1 fusion: one target or two?  Level 2 fusion:
assessment (who are they? what are they doing?).  Level 3 fusion: impact
assessment (how will their actions affect me or will my actions affect
them?). Level 4 fusion: process refinement (how can i get more info?).
Show information, whether it's certain, show how to get more info.

2. Understanding uncertainty.  Knowing what you're not sure of.  
How do we represent what is unknown?  In experiment: humans were able
to classify ISAR imagery around 56% accuracy.  Gave them simulated
classifiers with 70%, 80%, or 90% confidence level.  Humans did a little
better than on their own, but worse than the classifier would have done
on its own.  Classifier had to be 90% confident before humans would
begin to trust it.

3. Using the right metaphors.  In most cases we currently use maps
and the WIMP desktop metaphors.  Suggestions: TV metaphor (production,
storyline, acting, personalities, director) -- not compelling.  Is
there another metaphor that will cause everything to fall into place?

4. Applying decision science.  Naturalistic decision making suggests
that people are not classical, rational decision-makers; people tend
to recognize a set of features that remind them of their own experiences,
and through recognition they recall the relevant solution from memory.
So people don't consider all options; rather they choose an option first,
run a simulation in their head, and if it works they decide the solution
is sufficient.  Humans have information processing biases, e.g. selective
perception, framing bias, etc.

5. Visualization for asymmetric warfare.  Where is the battlespace?

Three levels of picture: strategic (whole picture), operations (theater),
tactical (real-time).

Tailored VIsualizations at CPOF
-------------------------------

William Wright (for Ward Page)

Increase speed and quality of command decisions; smaller,
more agile process.  Tailored visualizations.  Blobs.

Land force visualization uses icons and a visual language from the 17th c.