Office of Naval Research
Solicitation For Research Proposals: Tactical Picture Agent
Purpose:
This announcement solicits submission of research proposals for a
5-year research initiative on the topic of intelligent software agents
that can dramatically improve future Naval capabilities for achieving
common, consistent, coherent Tactical Picture among Naval decision-makers at
sea
and ashore.
Background:
The ability of Naval commanders to conduct effective, coordinated
at-sea operations is fundamentally determined by their ability to
access and utilize the right information at the right time. Naval
commanders need to be able to understand as accurately as possible all
important factors regarding their own situation, including but not
limited to the
disposition of Allied, Neutral, and non-combatant entities,
geographical, oceanographic,
and meteorological characteristics in the areas of
interest, and of course, the disposition of an adversary if one exists.
The totality of the information relevant to a Naval commander is
referred to as the Tactical Picture. The Navy's objective is
to provide commanders with the most accurate, comprehensive, and
consistent tactical picture that technology will allow, under the
presumption that improved tactical picture will improve the ability of
the Navy to conduct operations effectively and efficiently.
Future Naval ships and other platforms will have the capability to
draw upon varied and voluminous sources of data that might potentially
have relevance to peace-time, crisis, or war-time operations. In
addition to conventional sources of tactical information such as Naval
text
messages and tactical data such as radar and sonar tracks, commanders
will have access to the burgeoning "infosphere" of publicly-accessible
and proprietary data repositories of weather maps, satellite images,
on-line newspapers, intelligence analyses, and newsgroups that the
world's interconnected networks are rapidly making available. Future
Naval and National surveillance and data acquisition capabilities will
be able to collect enormous quantities of information regarding
conditions in Naval operational theaters in the open ocean and in the
world's littoral zones. This information, together with the vast
information resources linked to the world's highly interconnected
computer networks, represents a data mining, filtering, and display
problem of substantial magnitude. It is not the objective of this initiative
to devise the technology for amassing tactical information, but rather to
maximize the utility of existing information to Naval commanders.
Long-Term Vision:
A key to effective utilization of the "infosphere" by Naval
commanders is to provide them with automated aids able to discover,
retrieve, and display information relevant to the operations being
conducted by the commander's platform (e.g., ship), regardless of
whether the
commander is aware of the existence of such information or able to
explicitly direct that such information be retrieved from specific
sources. Ideally, a smooth, adaptive man-machine interface would
facilitate discovery and acquisition of potentially useful
information, filter information for relevance and importance in the
context of current or projected operations, and then portray the
resulting information in a rational and understandable fashion.
Approach:
There are many technical hurdles to be overcome so as to be able to
achieve the long-term vision for Naval Tactical Picture. The premise
of this research initiative is that intelligent software agents
represent a potentially fruitful approach for addressing problems of
discovery, relevance filtering, and display of information relevant to
construction of Naval Tactical Picture, and user interactive capability for
efficient and effective use of this information.
Objective:
The Tactical Picture Agent initiative is an anticipated 5-year basic
and exploratory research effort intended to foster the development of
technology that can eventually dramatically improve tactical picture
on-board Naval platforms. Research is needed in the following main
areas, which are elaborated below:
- Modeling of users and tasks so that intelligent software can
decide what to search for and how to integrate search results.
-
Agents that locate and filter multi-media
information appropriate for a particular
user and task.
-
Methods of portraying information that are appropriate for the task
and for the environment in which it is carried out.
1. Modeling Users and Tasks
What is important for one task and Naval commander at a given time may
be irrelevant to another, and what is important at one time may be
irrelevant at another. To be able to filter only the relevant
information to present to a user, a model of the role of that person
and of the task being carried out is essential. The models are
interposed between the user-interface and information sources to
filter the available information according to the needs of the task
and the user.
Associated with each task or each person is an agent or set of agents.
The agents maintain the models and the current state of the task.
They are responsible for determining what information is needed at
each step of the task. The agents initiate searches, which may be
performed by other information filtering agents, to get the
information. When information is returned, it must be appropriately
combined with other information and displayed to the user. Depending
on the task, it may also be necessary to communicate with other agents
or users to request or provide information or to collaborate on a
task. It then becomes necessary to provide a means of ensuring that
the common information and task models are consistent.
2. Information Location and Filtering
Given an explicit or implicit request for information, the purpose of
information location and filtering is to select all and only that
information that is
relevant and reduce it to a manageable and understandable response.
Information relevant to a given task is a small subset of the total
information in the "infosphere" accessible at any given time.
Relevance is also a function of time, in that different items are
important depending on the task and conditions in the environment.
For Naval Tactical Picture, a wide variety of multi-media information types
will
be available, including text, pictures, maps, sensor information, etc.
This information is likely to be represented in a non-uniform way
using a wide range of formats. There will be different costs
associated with information from different sources, and some of the
sources might be more reliable than others. It is important for
intelligent agents that search and retrieve relevant information to do
so in a manner that will conserve scarce network bandwidth and
processing capabilities.
Because of time constraints and the heterogeneity of the information
and of the repositories, it will generally not be possible to perform
exhaustive searches throughout the "Naval Internet", and it will most
likely be necessary to conduct multiple searches in parallel. This
leads to the possibility of multiple agents that cooperate or are
loosely coordinated, and conduct separate searches for information.
Research is needed into controlling multiple agents, negotiation among
agents, maintaining consistency, dealing with cost and reliability
issues of information sources, efficiently searching for and
transmitting information, minimizing redundancy,
and ensuring adequate coverage of the
information sources. Agents will operate in a heterogeneous networked
environment, with different protocols and database systems, and a wide
range of computer systems. This may require research into
transportable agents and common database access methods.
3. Portraying the Information
The problem of displaying information to the user is complicated by
the wide range of display devices that may be available and the
requirement that task execution not be adversely impacted.
Multi-media information may be retrieved in many forms, some of which may
not be
suitable for display on available devices. For example, some devices
may be unable to display pictures, while others may not provide audio.
Determining the most appropriate methods of display, translating
between one representation and another, and integrating new
information with the existing task are all challenging research
problems. Both the human-computer interface and the quality of
information are important (a good summary badly presented and a bad
summary well presented are both of limited value). Research is needed
to address the generation of summaries and the determination of how to
present them given the available output devices.
Nature of the Research Initiative:
The Tactical Picture Agent initiative is intended to develop
demonstrable technology prototypes operating in the context of
simulated Naval data sources by the end of 5 years. During the first
few years, the intention is to pursue research efforts that
concentrate on basic investigations into powerful and novel approaches
involving intelligent software agents for discovery, retrieval,
filtering, and display of tactical picture information. During the
latter years, research efforts will be encouraged to focus on
exploratory technology development and demonstration. Researchers
will be encouraged to work closely with Naval laboratories in
demonstrating their technical concepts. Total funding for this
initiative is anticipated to average approximately $3.5M per year.
Subject to anticipated funds availability, multiple awards will be
made.
Instructions for Proposal Submission:
Basic research and exploratory development proposals that address one
or more of the technical issues outlined above are solicited.
Proposals from universities, research institutes, corporations, and
Government entities with innovative ideas are encouraged. Proposals
should not exceed a total of 30 pages, single-spaced, 10-point type.
Proposals should follow the General Proposal Preparation instructions
found in the ONR Guide to
Programs, and should specifically include an executive summary
that summarizes the proposed technical approach, a clear description
of the specific problem or issues to be researched, a comprehensive
description of the novel technical approach being proposed,
relationship of the proposed work to other work reported in the
literature, previous or on-going related work conducted by the
proposer, technical resumes of key investigators, and budget in
sufficient detail to permit assessment of proposed costs.
Supplementary material such as videotapes or CD-ROMs may be included.
Proposals should be prepared with a start date of 1 June 1995 and a
period of performance of no longer than 3 years. Five copies of
proposals must be submitted in hard-copy format (no faxes or e-mail)
to the following address by 1630 hours on Friday, April 7,
1995:
Office of Naval Research
ATTN: Tactical Picture ACI
ONR Code 311
Ballston Centre Tower 1
800 N. Quincy St.
Arlington, VA 22217-5660
Awards under the Tactical Picture Initiative will be made subject to the
conditions of ONR's Broad Agency Announcement published in the Commerce
Business Daily on 16 June 1994.
Technical and administrative questions regarding this research
initiative may be directed to Dr. Michael O. Shneier,
(703) 696-4303,
shneiem@onrhq.onr.navy.mil.
Prospective proposers should refer to this homepage occasionally to check
for clarifications or modifications that may occur before proposals are due.