|
|
|
Introduction
To respond to an ever-changing,
uncertain environment, human commanders must maintain a general
awareness of the battlespace, and yet focus on relevant contextual
information when making decisions, without distraction from peripheral
events. Decision cycles have shortened, available information
has expanded, and missions have become increasingly variegated.
The repertoire of traditional battle functions has expanded to
include new types of operations, such as ad hoc responses
to terrorism, response to biochemical threats, and operations
other than those of war and coalition operations. To meet these
new challenges, the commander needs enhanced decision support.
Yet providing this automation becomes increasingly difficult as
uncertainty increases.
Furthermore, while
the volume of raw information available for command decisions
at all echelons is rapidly increasing, its coordination and dissemination
as useful information becomes far more difficult, leading to the
problem of "data overload and information starvation."
Data is often fragmented, multi-modal, uncertain, and distributed
across disparate sources.
The next generation
of battlefield information systems must meet the twin challenges
of scaling up to accommodate the explosion of cheap ubiquitous
sensors, while extending access to increasingly heterogeneous
information sources, from the Air Force's own legacy systems to
those of other cooperating services and nations. These new sources
of information must not only be accessed; they must be converted
from mere data into sources of usable, actionable knowledge.
To
address these issues in high-level information fusion, we are
conducting a multidisciplinary research effort involving computer
scientists, engineers and cognitive psychologists, from Carnegie
Mellon University, the University of Pittsburgh, the Munitions
Directorate of Air Force Research Laboratory (MN/AFRL), Rome Labs,
and Northrup Grumman. We will combine our various expertise to
develop the next generation of information fusion systems.
The problems
that must be addressed in order to build these systems are:
- Integration
of heterogeneous information sources both internal and external
to the Air Force (e.g. intelligence reports)
- Adaptation/Customization
to rapidly changing threats and missions.
- Automation
of information selection and dissemination to provide the right
information to the right decision maker at the right time, while
also keeping pace with the shrinking decision cycle.
- Scalability
to accommodate ubiquitous sensing and other expanding information
sources.
To meet these
challenges a next generation information system must:
- Be Modular
and flexible enough to configure itself without explicit guidance,
so that it can be deployed and used immediately.
- Be Open
enough to accommodate dynamically changing and heterogeneous
information sources, legacy systems, and various levels of fused
information.
- Support
automation to direct attention; and filter, and distribute information
to warfighters at all levels.
- Support
technology for quality of information pedigrees.
- Support
information fusion across levels, maintaining coherent pictures
from varied perspectives and levels of aggregation.
For
future information processing and planning systems, an in-depth
understanding of the cognitive processes of the user being aided
must be understood, coupled with innovative approaches for real-time
information fusion at all levels, including multimedia and multi-modal
information from disparate and distributed sources that include
enormous amounts of uncertainty and noise. Such cognitively congruent
systems will provide an intuitively understandable common operational
picture for enhanced situation assessment and battle management,
along with planning guidance and monitoring functions in the uncertain
and quickly evolving battlespace. Our overall research hypothesis
is that the way to address the issues in information fusion at
different levels is through adaptive and self-organizing collections
of Intelligent Agents who also possess models for discriminating
and communicating situational distinctions salient to humans and
the current mission.
Our
multidisciplinary and multi-institutional research project research
will help translate information superiority into decision superiority,
i.e., to make information into knowledge, in order to rapidly
arrive at better decisions than adversaries can respond to.
Robotics
Institute Project Page
|
|
