PROPOSER INFORMATION PAMPHLET (PIP)

IMAGE UNDERSTANDING FOR BATTLEFIELD AWARENESS (IUBA)

Broad Agency Announcement (BAA 96-14)

Defense Advanced Research Projects Agency (DARPA)

The information provided in this Proposer Information Pamphlet (PIP), in addition to that provided in the Commerce Business Daily (CBD) Announcement, BAA 96-14, constitutes a Broad Agency Announcement as contemplated in FAR 6.102 (d)(2)(i).

Technical POC: Dr. Thomas M. Strat, DARPA/ISO, [email protected], FAX: (703) 696-2201

Contractual POC: Ms. Algeria Tate, DARPA/CMO

SUMMARY OF IMPORTANT DATES

24 May 1996 - Proposal Abstract Due Date

23 July 1996 - Proposal Due Date

23 July 1996 - BAA 96-14 Closes

September 1996 Anticipated Evaluation Completion/Notification

2 January 1997 - Contract Award

Order of Topics

I. INTRODUCTION

A. Purpose

B. Objectives

C. Background

D. Approach

1. Category A: Integrated Feasibility Demonstrations (IFD)

2. Category B: Focused Research Effort (FRE)

E. General Guidance for Generating Proposals

II. DESCRIPTION OF TECHNICAL AREAS

A. Technical Area 1: Image Exploitation (IE)

1. Objective

2. Application Scenario -- Force Monitoring

3. Background

4. Enabling Technologies and Key Problems to be Addressed

5. Experimental Support

6. Guidance Specific to Categories

B. Technical Area 2: Automatic Population of Geospatial Databases (APGD)

1. Objective

2. Application Scenarios

3. Background

4. Enabling Technologies and Key Problems to be Addressed

5. Experimental Support

6. Guidance Specific to Categories

C. Technical Area 3: Video Surveillance and Monitoring (VSAM)

1. Objective

2. Application Scenarios

3. Background

4. Enabling Technologies and Key Problems to be Addressed

5. Technologies Not of Interest in This Technical Area

6. Experimental Support

7. Guidance Specific to Categories

8. Guidance for IFD and FRE Collaboration in the VSAM Program

D. Technical Area 4: Automatic Target Recognition -University Research Initiative (ATR-URI)

1. Objective

2. Background

3. Experimental Support

4. Guidance for Generating ATR-URI Proposals

III. GENERAL INFORMATION

A. Abstract Preparation and Delivery

B. Proposal Preparation and Delivery

1. VOLUME I: Technical Proposal

2. VOLUME II: Cost Proposal

C. PROPOSAL DELIVERY

IV. EVALUATION CRITERIA

V. TECHNICAL AND ADMINISTRATIVE INFORMATION

VI. ADDITIONAL RESOURCES: EXPERIMENTAL DATA

VII. OTHER IMPORTANT ADMINISTRATIVE INFORMATION

I. INTRODUCTION

A. Purpose

The Defense Advanced Research Projects Agency (DARPA) is soliciting proposals for research and new technology development related to Image Understanding for Battlefield Awareness (IUBA). Comprehensive battlefield awareness will provide US military forces a decisive advantage in future military actions--Image Understanding (IU) is one of the key technologies that will fulfill the requirements for battlefield awareness. The future battlefield is characterized by a rapidly expanding suite of sensors and sensing modalities collecting unprecedented volumes of imagery from a mixture of ground, air, and space-borne platforms. Image understanding techniques are needed to extract the information needed by military forces from this data-rich environment.

Proposed research should investigate and demonstrate innovative approaches and techniques that lead to or enable revolutionary advances in the state-of-the-art. Specifically excluded is research that primarily results in evolutionary improvement to the existing state of practice.

B. Objectives

The DARPA IU Program provides the technology base for DARPA-sponsored programs requiring image understanding techniques. The objective of the IU Program is to develop and demonstrate radically new, automated techniques for extracting relevant information from all modalities of imagery. The IU Program supports advanced research and development that will lead to revolutionary technologies and tools for use by other DARPA and DoD programs. Consequently, program results must demonstrate the potential to support the image understanding needs of existing and forthcoming DARPA and/or Department of Defense (DoD) programs.

The focus of the IU Program is to enable increased battlefield awareness through the

development of innovative computer vision techniques and novel employment concepts. This BAA solicits proposals for coherent programs of research directed toward this goal. Proposals are sought that offer focused research on enabling technologies, and integrated feasibility demonstrations to exhibit more advanced and more reliable IU algorithms than are currently possible.

C. Background

The DARPA IU Program has developed new and innovative IU techniques for several decades. Much of this research has matured to the point where its incorporation into fieldable systems is now at hand, while further research is needed to extend and deepen the IU capabilities that are currently limited.

The unreliability of IU algorithms has severely limited more widespread adoption of IU systems in many domains. Techniques and approaches that promise significant advances in overall system reliability are of great interest. The reliability barrier is not likely to be broken through incremental improvements to known algorithms. Instead, research should seek methods to build reliable IU systems from unreliable components, or should pose different IU tasks that are both DoD relevant and can be solved reliably.

While processing speed is a critical issue for many applications of IU, acceleration of existing IU technology is not desired in response to this BAA. Priority must be given to developing IU-based systems that are sufficiently reliable for their intended tasks.

Proper evaluation of IU developments is a matter of scientific integrity and is the responsibility of every researcher. However, the lack of an agreed upon theory of vision and the extreme variability which IU algorithms are expected to handle make it infeasible to perform a complete evaluation in every case. Direct comparisons between two algorithms are seldom possible because independently developed IU algorithms are seldom designed to function in the same domain under the same circumstances. Nevertheless, it is incumbent upon every offeror to conduct experimental evaluations that measure the extent and limits of his contribution within the bounds of pragmatic reality. These experimental evaluations should be conducted using rich, varied data sets. Offerors are expected to devise metrics appropriate to the task, set performance goals in terms of those metrics, and perform rigorous scientific evaluation at regular intervals consistent with the data available. While imperfect, such evaluation provides a valuable measure of the progress of the field, and can serve to focus further research on shortcomings revealed in the process. The DARPA IU community must ultimately demonstrate the value of its efforts in a military context. Technology transfer occurs in a variety of ways: published papers, trained students, algorithm sharing, incorporating algorithms into the Image Understanding Environment (IUE), commercial products, fielded systems, etc. Offerors must state how their results will be transferred to others who can make use of them, bearing in mind that some transfer paths are stronger than others.

Additional background information on the DARPA IU program can be found at URL:

http://www.hokie.bs1.prc.com/baa96_14

Recent technical papers documenting progress in image understanding may be found in the Proceedings of the DARPA IU Workshops of November 1994 and February 1996 (Morgan Kaufmann Publishers, San Francisco; Web site: http://www.MKP.COM; e-mail: <[email protected]>; phone 415 392-2665).

D. Approach

The technologies targeted by IUBA should be on the critical path to creating an increased level of battlefield awareness using imagery and related data that is now available or can reasonably be expected to be available to the military in the near future. Proposed research and development efforts should investigate innovative approaches to overcoming technical barriers and enabling revolutionary advances.

Proposals are sought that address the following Technical Areas:

1) Image Exploitation (IE)

2) Automatic Population of Geospatial Databases (APGD)

3) Video Surveillance and Monitoring (VSAM)

4) Automatic Target Recognition - University Research Initiative (ATR-URI)

In Technical Areas 1 - 3, awards will be made in two categories: (A) Integrated Feasibility Demonstrations (IFDs) and (B) Focused Research Efforts (FREs). Only FREs are solicited for Technical Area 4.

In Technical Area 4, Proposals may be submitted only by U.S. Institutions of higher education (other than federal government) with degree-granting programs in science and/or engineering, or by consortia led by such institutions.

A total of approximately $10 million is available for funding multiple efforts in accordance with this BAA beginning January 1, 1997. Additional funding for FY98 and beyond is subject to availability of funds. The sizes of awards and periods of performance of efforts will vary according to the type of effort undertaken (see below).

1. Category A: Integrated Feasibility Demonstrations (IFD)

Of interest are efforts to build relevant, application-ready, software-based, end-to-end, prototype systems that provide unique or novel embedded IU capabilities in one of the Technical Areas (1-3) listed above, and can be demonstrated and evaluated within a reusable, instrumented, software-based infrastructure based on emergent object-oriented industry standards. IFDs are laboratory demonstrations to be carried out at the contractor's facility. IFD proposals must address an IU application as defined below in the description of Technical Areas, implement the software technologies in a quasi-operational environment, measure the effectiveness of the IU tools and techniques, and show an improvement in the ability to provide the commander an enhanced awareness of the battlefield. Innovative approaches to facilitate the transition of those research results into other application programs sponsored by DARPA or other DoD organizations are desired.

Proposals for an IFD must show a clear understanding of the battlefield awareness process being addressed and the appropriate role of advanced IU technologies in improving that process. The IFD proposal must describe the concept of operations for the end-to-end system and its embedded technologies, the suite of technologies to be developed, the system architecture that will incorporate those technologies, and a plan for periodic demonstrations over the project life. Proposals must carefully describe the anticipated capabilities and limitations of the proposed technology, and include an evaluation plan to ascertain the limits of the implemented system. The proposals must describe how technology developed independently, such as an FRE under Category B, described below, could be integrated into demonstrations during the course of the program.

The proposal must include a discussion of how software and algorithms will be exchanged among organizations participating in the IFD.

The IFD proposal may include focused research to be conducted by the IFD contractor. The IFD is intended to motivate the creation of new IU technology, not just integrate existing techniques.

The IFD contractor in the IE and APGD Technical Areas must have capabilities to store and process classified imagery and other material. A security plan must be provided, according to the instructions in the Section III, General Information, Part B, Volume I, paragraph N.

All awardees must attend and present their projects at the annual Image Understanding Workshops, which are held each year at alternate coasts (East Coast in 1997, West Coast in 1998)

IFD contractors must:

Proposals should be for a 24-month performance base, plus options for tasks to be performed during months 25 - 60. The IFD must be completed within the base 24-month period. Optional effort shall offer the following:

Proposals for IFDs should not exceed $1.5 million per year. Offerors should propose efforts consistent with the technical challenge and the resources needed to address the challenge.

2. Category B: Focused Research Effort (FRE)

Of interest are new research efforts that emphasize the search for revolutionary advances in the theoretical foundations of IU and are focused on one of the Technical Areas under consideration in this BAA. Proposed efforts should complement previous accomplishments from the IU Program and lead to the development of a testable demonstration within 24 months. The demonstration will provide an environment to prove out the concept and will be the basis for committing further resources required for more complete investigation and testing. Even though the emphasis for FREs is on innovative research, the effort must address an end-to-end application, or be demonstrated as a component of an existing end-to-end system. Demonstrations of results are to occur at regular intervals as appropriate to research plans (minimum every 12 months). Examples of topics for FREs are given in the description of each Technical Area below.

FRE proposals must describe a research and development program that includes: technology description and comparison with alternative approaches, innovative ideas, anticipated impact of research on battlefield awareness, a demonstration plan with specific deliverables, evaluation metrics and success criteria to assess the impact of the developed technologies, and an approach for interaction with other IUBA contractors.

The FRE contractors will attend and actively participate in the two semi-annual workshops coordinated by the IFD contractor. These meetings will generally be held as appropriate at the IFD site, at field experiment and demonstration locations, or in conjunction with other appropriate conferences or program workshops.

The FRE developers will, at minimum, conduct one experiment/demonstration of their research every year. They may choose to do this at the IFD-conducted workshops or at their own sites, as appropriate. Additional periodic demonstrations of emerging systems are encouraged, as long as they do not strongly impact research progress.

It is anticipated that most FREs will be unclassified and will not require access to classified data. However, classified efforts are not excluded.

All awardees must attend and present their projects at the annual Image Understanding Workshop, which is held each year at a location within the U.S.

FRE contractors must:

Proposals should be for a 24-month performance base, plus options for tasks to be performed during months 25-60. Optional effort shall offer the following:

technology transfer activities to integrate the techniques into an IFD or other DoD program.

Multiple FRE awards are anticipated in each Technical Area. Proposals for research efforts in the range of $100K - $300K per year are desired.

E. General Guidance for Generating Proposals

Collaboration is essential between the IFD contractor and the FRE contractors in each Technical Area. However, the IFD contractor should not depend on specific results from the FRE contractors.

The IFD contractor must make their imagery and collateral data available to the focused research contractors in their technical area. The Government will augment existing datasets with additional imagery (including classified imagery) during the period of performance and will actively seek out new data sets and make them available to the IUBA contractors. It should be noted, however, that the individual contractors are ultimately responsible for acquiring imagery and other data needed to perform proposed experiments and evaluation. After contract award, the IFD contractors will be responsible for reproduction of Government-provided imagery, producing accurate camera model information, and distributing imagery and associated metadata to the FRE contractors.

Reusable IU technology is desired to accelerate progress in the field. Offerors are encouraged, wherever practical, to modularize and encapsulate the most significant algorithms to be developed, and identify them as contract deliverables.

For purposes of exchange of data and algorithms within the IU community the use of the RADIUS Common Development Environment (RCDE), Image Understanding Environment (IUE) and/or Khoros is encouraged if appropriate. In the event that the IUE or RCDE is not used, the ability to interchange data with the RCDE and/or IUE is required.

Additional information on RCDE can be found at URL:

(this URL is "Under Construction")

Additional information on IUE can be found at URL:

http://www.aai.com/AAI/IUE.html

The IUE provides one particularly effective mechanism for achieving technology transfer. Contributions of well-written software to the IUE libraries are desired. An evaluation plan is required for all proposals. The proposed work must include an approach to design and testing which permits rigorous scientific evaluation. The following elements must be addressed in the plan:

Testing must explore the limits of competence in terms of those metrics.

All proposals must be unclassified. Assume a January 1997 start.

II. DESCRIPTION OF TECHNICAL AREAS

Proposals are sought that address one of the following technical areas. Proposals should indicate whether they offer (A) Integrated Feasibility Demonstration (IFD) or (B) Focused Research Effort (FRE).

A. Technical Area 1: Image Exploitation (IE)

1. Objective

The objective of the IE Technical Area is to push the development of IU technology so that IU systems can provide comprehensive support to tactical imagery analysts in identifying and tracking military forces and force movements. Because the volume of imagery expected from Unmanned Aerial Vehicles (UAVs) and national collection systems is so great compared to planned image analyst (IA) staffing levels, truly revolutionary change in image exploitation is needed, not incremental improvement in IA productivity. A secondary objective of this Technical Area is to focus the IU research community on the challenges presented by the need to automatically extract salient information from imagery of multiple modalities and from multiple images acquired over time.

2. Application Scenario -- Force Monitoring

In our vision of the future, image exploitation will be performed not by searching large tracts of terrain for significant forces and activities, but by monitoring and tracking the movements of all military equipment within a large, but limited, geographic area. This vision will be enabled by maintaining a vast repository of information about potential battlegrounds, such that image interpretation can be conducted in comparison to prior knowledge. Vast quantities of imagery will be collected routinely, and used to update the repository over as wide an area as is technically feasible. Image exploitation, then, could be performed with the aid of a wealth of data that is not available to today's IA.

For example, in the near-term, the IU-assisted IA could prepare to track ground movement of important vehicles in a restricted geographical area (say a 5 km x 5 km battlefield, an airfield, a garrison, a bivouac site, or a road junction), by extracting salient information from imagery of that area that is routinely collected for whatever purpose. This task can be compared in some ways to air traffic control surveillance except it is focused on ground vehicles. As hostilities near, the IE system will use imagery from UAVs, which are operating in a flight pattern that provides frequent coverage (every hour or every minute). Near-real time intelligence reports are issued by the IA with the assistance of highly reliable IU tools.

The repository of prior information, a dynamic database, is used for many purposes: to focus the attention of the analyst and his automated tools, to inform the system about what to look for, to choose the most appropriate algorithms, to set their parameters, to reduce the potential of false alarms, and to evaluate the validity of the results. The database should contain whatever is needed to perform these tasks reliably -- likely possibilities include cartographic products, prior imagery, image features, statistics from previous use of algorithms, two- or three-dimensional models of cultural features, etc.

The intended user of the IE system is a tactical IA who is monitoring force units that move, then sit (for hours or days or months), then move. Large quantities of imagery from UAVs -- primarily synthetic aperture radar (SAR), some electro-optical (EO) and infrared (IR) -- must be analyzed by the IA with IU assistance.

Examples of tasks to be carried out include:

The key research issues to be addressed in this technical area are:

  1. What information should be stored to assist with the automated interpretation of future imagery?
  2. How can IU systems be designed to use that information effectively?

3. Background

Research in this technical area is intended to support any or all phases of image analysis. Because of the severe burden that the extraordinary volume of imagery collected from UAVs places on the IAs, automated techniques to help analyze UAV imagery is of particular interest. For further information on the High Altitude and Endurance (HAE) UAVs, see URL: http://www.arpa.mil/asto/tier3.html. Approaches that can also deal with other sources of imagery, including national technical means, are also desired.

There have been several research and development (R&D) programs that have been concerned with IU support to the IA. Semi-Automated IMINT Processing (SAIP) and Research and Development for Image Understanding Systems (RADIUS) are notable examples of such programs. Descriptions of RADIUS may be found in the Proceedings of the ARPA IU Workshops of November 1994 and February 1996 (Morgan Kaufmann Publishers, San Francisco; Web site: http://www.MKP.COM; e-mail: <[email protected]>; phone 415 392-2665). Descriptions of these programs are also available on the World Wide Web (WWW) under "Additional Proposer Information" at URL:

http;//www.darpa.mil/baa/#iso

The RADIUS project has shown considerable progress in national (or strategic) imagery exploitation applications. Efforts in this Technical Area are intended to build on RADIUS results and complement them. It is anticipated that such projects will eventually be integrated into RADIUS-like exploitation environments.

Contracts in support of this technical area are to develop and demonstrate automated IU tools to assist the IA. Such tools should focus on IA tasks and techniques beyond those currently planned for incorporation into SAIP. While augmentation of SAIP would be a desirable objective, offerors should not be constrained by the current processing timelines, imagery sources, or concepts of operations embraced by SAIP. The IFD for IE should realize a vision for the next generation of automated image analysis workstations.

The SAIP Advanced Concept Technology Demonstration (ACTD) program is a DoD initiative to develop and field a set of integrated exploitation tools that will greatly increase the ability of IAs to exploit large amounts of imagery generated by sensors on the HAE UAVs and the U-2R (ASARS-2 and upgrades to the ASARS-2). It will enable the IA to more accurately and rapidly perform both target search operations (for time critical targets such as missile order of battle targets and company-size groupings of ground order of battle targets) and point target exploitation (of airfields, garrisons, and other critical military sites within the theater). The goal of the SAIP system is to provide the IA the necessary tools to analyze the imagery and generate reports to commanders and intelligence planners within a few minutes after data collection is complete. Additional information on SAIP can be found under "Additional Proposer Information" at URL:

http://www.darpa.mil/baa/#iso

4. Enabling Technologies and Key Problems to be Addressed

Revolutionary improvements in IU technologies are expected to be derived from the more complete use of context and stored knowledge. Other research topics include multi-source IU, interactive IU, and IU augmentation of automatic target recognition (ATR) algorithms. These topics are briefly described below as an aid to bidders. However, the technologies discussed in this section are neither exhaustive nor required. Offerors are encouraged to propose any IU research topic which promises a major advance in image exploitation.

The use of multi-source imagery is important, including EO, SAR, IR and other imaging modalities. IU applied to SAR imagery is of particular interest, because of the all-weather, day/night capabilities of SAR. A related area of interest is the use of multiple-source images from last month, yesterday, or a few hours ago, to help understand a new SAR image.

Context

IU performance can be improved by using spatial, temporal, physical, and linguistic context. The RADIUS program produced significant IU accomplishments making use of static structural and cartographic context in site models. This Technical Area is concerned with extending the use of context pioneered by RADIUS by combining the physics and geometry of computer vision with dynamic models and linguistic information.

The use of temporal context as an aid to IU is of great interest, as is the choice of an appropriate representation for temporal knowledge used by IU. Offerors should consider approaches to improving performance by adding temporal models (or process models or dynamic models) of entities being observed.

Physical models can provide thermal, material, radiometric, and spectral properties of objects under observation and the surrounding environment. IU algorithms must make effective use of such models.

Additional context useful for IU might arise from integrating IU with language understanding. IAs use much information that is available in textual form (previous reports and descriptions) to provide context for their analysis. This IUBA BAA will not support new research in language understanding, but the integration of existing capabilities with IU systems might be helpful to an image exploitation system.

The use of multiple knowledge sources can provide context for IU. Such information includes terrain, previous imagery, maps, site models, object models, sensor models, textual collateral data, and others. Information in these sources can support knowledge-based vision.

Knowledge Representation

This IUBA BAA is not aimed at knowledge representation research or symbolic reasoning research. However, the use of knowledge representation or symbolic reasoning results within image understanding might provide important advantages. The adaptation and integration of knowledge representation and uncertain reasoning techniques developed by the artificial intelligence (AI) research community may be appropriate.

A related research area is multi-dimensional databases, which include space and time. This BAA will not support new research in database retrieval, but the integration of existing capabilities with IU systems might be applicable.

Multispectral and SAR IU

Research is sought that identifies and solves the unique IU issues associated with SAR. Approaches to exploitation of SAR images of urban scenes are of interest, as is IU which analyzes the information in multiple-source images. A combination of the spatial representation and reasoning techniques developed by the IU community with the spectral processing techniques is desired to advance the exploitation of multispectral imagery.

Interactive IU

It is anticipated that IU algorithms will screen, filter, and prioritize images and then present selected images to the IA for analysis and decisions. Thus IU examines many images, the IA examines only a few. There are a number of research questions associated with how an IA effectively interacts with, controls, and guides the IU process. Are there interactive techniques to enhance the IU-to-IA interface? Are there interesting ways to portray what the IU thinks it is observing on the battlefield to IAs and to commanders? The main focus is on IU techniques, but interactive components necessary to achieve robust performance are of interest.

5. Experimental Support

A data set of sample imagery is available to bidders. The data set consists of SAR and EO imagery. Use of these data sets to illustrate current capabilities in the proposals is encouraged. Section VI of this PIP provides instructions on how to obtain this data.

DARPA intends to augment this data set with additional imagery, including classified imagery, during the period of performance and will actively seek out new data sets and make them available to the IE contractors. It is the intent to make use, as much as possible, of imagery collected for the MSTAR and SAIP programs. During 1996, SAR imagery, both spotlight and stripmap, from various azimuths and elevations, will be collected of armored vehicles, trucks, and SCUDs in open terrain. These collections will include up to three different SAR sensors. Collections for 1997 and beyond are planned to fill out algorithm development needs and thus specifics have not yet been finalized. The goal is data sets representing a large variety of enemy activity including maneuver forces, command and control, air defenses, and surface-to-surface missiles with accompanying ground truth.

6. Guidance Specific to Categories

Category A: Integrated Feasibility Demonstrations (IFDs) for Image Exploitation (IE)

Goals: The goals of the IE IFD are:

  1. to identify and develop the algorithms and system components needed for an IU-based Force Monitoring system,
  2. to show that these components successfully work together in a laboratory environment,
  3. to perform rigorous scientific evaluation of the algorithms and system components in order to demonstrate their reliability, and
  4. to provide a basis for subsequent technology transfer to an operating prototype such as a DARPA Advanced Concept Technology Demonstration (ACTD).

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document.

The IFD offeror should propose an IE laboratory demonstration as early as possible that incorporates many of the capabilities needed for an operating prototype. Improved and upgraded laboratory demonstrations should be provided in the subsequent option years of the program. Clear and measurable milestones should be identified and presented in the proposal for both the base and option years. The proposal must describe an experimental environment, or testbed, to be used for evaluation and demonstration.

Proposed work must clearly show relevance to the imagery exploitation application. The laboratory demonstration should be aimed at eventual technology transfer to DoD programs.

The proposal should describe proposed innovative approaches to Force Monitoring, how the approach is expected to improve or change image exploitation, proposed extensions of the operating domain of selected algorithms, modifications planned to handle situations not encountered previously, and other proposed improvements.

The offeror's IFD approach must provide reliable system performance. Stand-alone IU is not desired. The proposal should address the exploitation problem using IU, site models, all sources of collateral and context, and constantly updated databases. Include multiple sources of information to provide context. Employ ATR algorithms where they are appropriate. The human IA should also be used (sparingly) for decisions and control.

An approach to hasty mapping should be included in the proposed IFD. Hasty mapping refers to the capabilities that permit visualization of the earth where maps and databases do not exist. The APGD Technical Area of this BAA addresses broad and general approaches to creation of geospatial databases. The IE IFD must include techniques for hasty mapping to be used when an image exploitation system is deployed to a new area where maps and databases of sufficient quality do not yet exist. Potential approaches to hasty mapping include: 1) ground classification using polarized SAR, and 2) generation of three-dimensional terrain and texture maps using multiple images, IFSAR, or shape from motion.

In addition to the hasty mapping efforts, the IFD contractor should also remain aware of the activities in the APGD Technical Area and cooperate with the APGD contractors as needed. It is expected that the techniques from the APGD Technical Area will be integrated with the results of the IE Technical Area in operational demonstrations in the out years.

It is anticipated that the IFD contractor must be able to store and process classified materials. However, much of the research work may be carried out by uncleared researchers using uncleared processing facilities. A research laboratory without classified facilities might selectively clear personnel and use the classified facility of another contractor on the team for classified experimentation and demonstrations. A security plan must be provided according to the instructions in Section III, General Information, Part B, Volume I, paragraph N.

Category B. Focused Research Efforts (FREs) for Image Exploitation (IE)

Goals: The goals of the IE FREs are:

  1. to expand the technical base supporting image exploitation with research that offers a promise for radical advances in image exploitation, and
  2. to particularly explore research ideas that offer a revolutionary change in the way an IA exploits reconnaissance imagery.

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document.

Proposals must indicate how major advances in image exploitation will be achieved if the research results are subsequently engineered and integrated into future systems.

As stated earlier in this PIP, a feasibility demonstration is required every 12 months. The demonstration must be embedded within an end-to-end image exploitation scenario.

B. Technical Area 2: Automatic Population of Geospatial Databases (APGD)

1. Objective

The objective of this Technical Area of the IUBA BAA is rapid and low-cost automatic population of geospatial databases, from a variety of imagery sources, to serve imagery exploitation and synthetic environment applications. IU methods are sought to automate the extraction of three-dimensional (3D) cartographic features and their attributes from imagery. Geospatial databases contain terrain, features, static and dynamic objects located accurately in a geographic coordinate system, along with associated attributes. Geometric and semantic resolution, precision, and confidence metrics are based on the requirements of one or more anticipated applications. A given database will serve several classes of users. Advanced IU techniques are needed to dramatically reduce the time required to construct geospatial databases.

2. Application Scenarios

Research in the APGD Technical Area will support two application scenarios:

DARPA's Synthetic Environments (SE) program, with emphasis on the rapid construction of virtual worlds (RCVW), and the military modeling and simulation (M&S) applications it supports:

These applications include: mission planning, tactical training, mission rehearsal, after-action review. Geospatial databases for the SE program may include one or more related areas of interest, large and small, including built-up areas and adjacent terrain, coastal areas and coastal seabeds, theaters of war and battlespaces including damaged areas, ground targets, and approach corridors.

Model-Supported Exploitation (MSE), with emphasis on site-specific tactical imagery exploitation applications:

These applications include: site monitoring; model-based change detection; battle damage assessment; and detection of camouflage, concealment, deception (CCD). A key area of research interest in APGD for MSE is the attribution of geospatial models derived from spectral and radiometric data to support activity analysis and temporal fingerprinting. Emphasis will be given to geospatial database updating as well as construction.

3. Background

The SE program is part of a major DoD initiative in Advanced Distributed Simulation (ADS). Key DoD participants in M&S include DARPA, the Defense Modeling and Simulation Office (DMSO), the Defense Mapping Agency (DMA), the Terrain Modeling Project Office (TMPO), the Simulation, Training, and Instrumentation Command (STRICOM), and the U.S. Army Topographic Engineering Center (TEC). The goal of the SE program is to improve the quality of entity level simulation of combat and environment representations. The intent is to provide synthetic battlespaces and environmental phenomenology that will add realism to the warfighting environment. The SE program has four technical objectives: develop new phenomenology at various levels of fidelity, develop models and protocols to represent and distribute synthetic environments to ADS participants, generate terrain and environmental databases at multiple resolutions, and the rapid construction of virtual worlds (RCVW). The APGD Technical Area of this BAA will complement the current technology development activities in RCVW at SRI International, GDE Systems, and Carnegie Mellon University, which are focused on nearer term virtual world construction goals. The results of the APGD research will be inserted in one or more software suites in DMA, TEC, TMPO, National Photographic Interpretation Center (NPIC) and their contractors that generate cartographic products and simulation runtime databases to support mission planning, tactical training, mission rehearsal, and after-action review.

Additional background on DoD M&S organizations can be found at URL:

http://www.dmso.mil/

http://www.tmpo.dma.gov:8001/

http://www.dma.gov/

http://cat.tec.army.mil

http://dis.pica.army.mil/

Additional background information on the SE program and RCVW can be found at URL:

http://svl.tec.army.mil/~bwalker/SE_HOME.html

Synthetic Environment Data Representation and Interchange Specification (SEDRIS) is a proposed format-independent data representation model for interchanging synthetic environment databases, including any combination of (but not limited to) terrain, ocean, atmosphere, 3D icons/models, features, topology, sound, textures, symbols, and special effects. The planned SEDRIS Application Program Interface (API) facilitates efficient and intuitive access to data stored according to the SEDRIS data representation model. The Level 0 API is designed for minimal but complete accessibility to SEDRIS compliant data. The Level 1 API should provide the capability to perform more abstract operations on SEDRIS compliant data. The SEDRIS data model proposes to incorporate existing standards for topology to establish data relationships and the DIGEST Feature and Attribute Coding Catalog (FACC) to define features, attributes, and attribute values. Additional background information on SEDRIS can be found at URL: http://ftp.sc.ist.ucf.edu/STDS/workgrps/se/land/SEDRIS//

SEDRIS is basing feature information on the DMA Vector Product Format (VPF) and its feature reference, the DIGEST Part 4, Feature and Attribute Coding Catalog. This catalog can be found at URL:

http://132.156.33.161/Engineer?DIGEST 1.2a/digest.html

Research in MSE is being pursued in RADIUS, a joint DARPA-Office of Research and Development (ORD) project. Its goal is to develop and adapt IU algorithms and systems that a) will increase the speed and accuracy with which 3D site models (for military installations, production plants, and other intelligence targets) can be constructed from multiple-source imagery and b) will provide advances in the speed, confidence, and range of analytical techniques with which IAs develop and report intelligence information derived from multiple-source imagery. RADIUS is addressing MSE as a new imagery analysis concept, with emphasis on the use of 3D site models in change detection, particularly for quick-look scenarios, site monitoring, detection and counting of mobile objects, and trend analysis. DARPA's initial support to RADIUS included the development of RCDE which is now in common use in the IU research communities and is the core of the RADIUS Testbed system. Current DARPA support to RADIUS includes five IU research laboratories in industry and academia in support of RADIUS objectives: Carnegie Mellon University, SRI International, the University of Maryland, the University of Massachusetts, and the University of Southern California. The results of this research have been and continue to be transferred to the RADIUS Testbeds at the Lockheed Martin Corporation (the RADIUS Testbed System prime contractor) and the National Exploitation Laboratory (NEL) at the National Photographic Interpretation Center (NPIC). The APGD Technical Area of this BAA will complement and extend the current DARPA technology development activities in RADIUS, with emphasis on geospatial database requirements in support of battlefield awareness and battle damage assessment.

4. Enabling Technologies and Key Problems to be Addressed

Proposals must address (but need not be limited to) one or more of the following enabling technologies in support of APGD. Illustrative issues are included in each case.

Geospatial database population from multiple sources:

Emphasis should be given to sensors that can supplement panchromatic EO images, including high resolution imaging SAR, high density Digital Elevation Models from IFSAR and other sources, and multi/hyperspectral sensors. Utility of DMA and other cartographic products in geospatial database population. Experimentation and analysis of effectiveness of data sources for specific database population tasks. Exploitation of combinations of heterogeneous imagery sources. Uses for secondary sources: intensification and/or confirmation. Photogrammetrically resolved camera models across multiple sources; registration, automatic selection of control features.

Utility of context:

Use of domain knowledge and other context pertinent to the task to reduce technical risk, improve quality, and/or improve efficiency through, for example, generation of more effective hypotheses, smarter algorithm selection and parameterization, reduction in search space, and efficient exploitation of what can be learned from past results. Exploitation of temporal context. Context-based IU architecture: explicit representation and storage of context; context sets and other methods for reducing high overhead of stored knowledge; interchange formats to encourage reuse of context and other knowledge on similar problems and in other environments. Evaluation of context-based IU: reporting results involving context-based IU experimental environments. Relationship to other modern paradigms, e.g., purposive vision.

Application of learning to IU:

Exploitation of site-specific or domain specific learned behavior to improve IU system performance: reducing the search space; avoiding repetition of false alarms. An IU-based system that is being used routinely for automatic population of even selected parts of a geospatial database provides an ideal basis for developing IU systems that learn from experience. Access to large amounts of imagery, large numbers of examples, and review by human operators, provide all the ingredients necessary for implementing a system that adapts to its ever-changing inputs. Because it is impossible to acquire sufficient training data in advance, or to hand-code sufficient knowledge to accommodate future unanticipated domains, a capability to learn through experience is essential.

Automatic attribution of modeled features:

High-level vision in the presence of context. Combinatorial evidence. Temporal analysis.

Integration of model components:

Integration/consistency of terrain, features, static and dynamic objects as a seamless whole (buildings lie on terrain surface, road edges meet terrain models, bridges meet roads, rivers flow downhill).

5. Experimental Support

The IFD contractor for the APGD Technical Area will be required to focus on a few sites for evaluation and demonstration. The primary site will be several areas within the Ft. Hood, Texas military reservation. Offerors proposing as the IFD contractor for the APGD Technical Area should use Ft. Hood data to illustrate their proposed approach. The Government will provide an initial Ft. Hood dataset to support generation of proposals in response to this BAA. See the instructions for requesting this imagery in the General Information section of the PIP.

A description of the Ft. Hood dataset and associated metadata is available at URL:

http://www.mbvlab.wpafb.af.mil/public/fthddata/

Offerors proposing an FRE in the APGD Technical area are also encouraged to use the Ft. Hood dataset for IU experimentation. Offerors may use other datasets that are representative of the kind of data sources used in operational environments.

Information about DMA products pertinent to this effort is available at URL:

http://www.dma.gov/products/products.html

The Government will augment the Ft. Hood dataset with additional imagery, including classified imagery, during the period of performance and will actively seek out new data sets and make them available to the APGD contractors, as government furnished information.

IFD and FRE proposals must provide an evaluation plan that includes a description of the datasets to be utilized and the steps that will be taken to acquire additional data required for evaluation, with emphasis on domain-specific and other context needed to support proposed IU capabilities.

6. Guidance Specific to Categories

Category A: Integrated Feasibility Demonstration (IFDs) for Automatic Population of Geospatial Databases (APGD)

Goals: The goals of the APGD IFD are:

  1. to demonstrate an end-to-end geospatial database population process: raw source imagery and context in (classified and unclassified) and exportable geospatial databases out,
  2. to develop robust, predictable automatic database population in a constrained, quasi-operational environment, with emphasis on rapid methods,
  3. to incorporate automated methods developed independently in demonstrations, and
  4. to transfer technology as embedded IU methods in synthetic environment database generation facilities and in softcopy imagery exploitation environments.

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document.

The proposal must describe an APGD testbed for demonstration and evaluation of automatic population of geospatial databases, based on reusable, instrumented, software infrastructure. Use off-the-shelf building blocks where feasible -- e.g., Geographic Information Systems (GISs), photogrammetric workstations, database management systems. The APGD testbed must be able to import algorithms and other results of FREs efficiently. The testbed should be designed as a quasi-operational environment to support one or both of the application scenarios. The rationale for design decisions must be provided.

The proposal must indicate how the offeror will develop, assess, and adapt IU and related technical capabilities for potential incorporation into an APGD Testbed. Identify gaps and deficiencies.

Give priority to seeking out available technology and systems in industry and academia. Propose technology development of automated methods where no available technology, algorithms, or systems are suitable for demonstration/evaluation within the application scenarios.

Collaboration is essential with the APGD FRE contractors, RCVW contractors, and RADIUS contractors on issues and problems. The offeror must insure that the FRE contractors in the APGD Technical Area have imagery data to support their experiments. As noted above, the IFD contractor will be responsible for reproduction and dissemination of Government-provided imagery, producing accurate camera model information, and distributing imagery and associated metadata to the FRE contractors. However, the offeror should not depend on specific results from the FRE contractors in proposing capabilities to be developed for the APGD testbed.

Research emphasis must be on automatic population techniques. Semi-automatic approaches may be proposed, but there must be an evolutionary path to automatic population.

Photogrammetric rigor is required in the offeror's approach to ensure accuracy of resulting geospatial databases.

The proposal should describe proposed performance improvements of selected automated model population capabilities, proposed extensions of the operating domain of selected algorithms, and modifications planned to handle situations not encountered previously.

A demonstration plan must be provided for one or both of the application scenarios. Demonstrations are required for the RCVW application; demonstrations for the MSE application are desirable.

Transferability of software to TEC, NEL, and other government laboratories is encouraged but not required. Participation by the offeror in SEDRIS workshops and other activities on interoperability standards is desired to facilitate technical coordination and technology transfer.

The IFD contractor in the APGD Technical Area must have capabilities to store and process classified imagery and other material. A security plan must be provided, according to the instructions in Section III, General Information, Part B, Volume I, paragraph N.

Category B: Focused Research Efforts (FREs) for Automatic Population of Geospatial Databases (APGD)

Goals: The goals of the APGD FREs are:

  1. to make revolutionary advances in IU needed to support APGD,
  2. to make advances in theoretical foundations of IU relevant to APGD, and
  3. to transfer of technology to the APGD testbed.

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document:

The proposal must address issues related to geospatial database population from one or more imagery sources as well as one or more additional enabling technologies/problems listed above.

Collaboration with the IFD contractor for APGD, the RCVW contractors, and RADIUS contractors on issues and problems is essential.

Research emphasis must be on automatic population techniques. Semi-automatic approaches may be proposed, but there must be an evolutionary path to automatic population. The offeror should build on previous and current efforts in RCVW and RADIUS.

A demonstration plan must be provided for the RCVW application scenario; a demonstration plan for the MSE application scenario is desirable.

It is anticipated that FREs will be unclassified and researchers will not require access to classified material. However, classified efforts are not excluded.

C. Technical Area 3: Video Surveillance and Monitoring (VSAM)

1. Objective

The objective of this Technical Area of the IUBA BAA is automated video understanding technology that may be used in future urban and battlefield surveillance and monitoring applications, where human visual monitoring is too costly, too dangerous, or otherwise impractical.

DARPA seeks one IFD and several FREs in this Technical Area. Both will focus on individual and combined VSAM technologies that may be used to address future battlefield and urban surveillance missions and tasks using video sensors on stationary, mobile or airborne platforms.

The general paradigm is that at some time in the future, one human operator at a remote host workstation may supervise many VSAM platforms to perform military battle area and urban surveillance operations. Processing technology will have advanced markedly to allow the processing of video streams of data, as opposed to individual video frames. The VSAM platforms (either stationary, moving on the ground, or airborne) will be networked, having multiple visual sensors and integral, autonomous VSAM processing and control. The platforms will have steerable, controllable sensors and may cooperate, sharing tasks or computation. The platforms will be capable of day/night operations and should be largely autonomous, notifying the operator of only salient information and engaging the operator minimally to alter platform operations or resolve platform problems. Accordingly, it is anticipated that a great deal of spatial and temporal knowledge and context (in addition to video/image processing) will be integral to VSAM performance.

2. Application Scenarios

A representative, but not exhaustive, list of battlefield and urban surveillance missions and tasks that may be addressed with such a system is given below. Offerors are encouraged to propose any missions, concepts of operation, and associated advanced video-understanding research that, in their judgment, shows significant promise and benefits for urban and battlefield surveillance.

Battlefield Surveillance and Monitoring Missions and Tasks:

Cooperative, multiple sensor and multiple platform monitoring of a battlefield area for incursion by enemy foot soldiers or mounted units.

Detection, tracking, identification, and geo-location of individual activities (soldiers or vehicles).

Activity understanding and discrimination: the recognition of activities from simple events (e.g. distinguishing infantry incursion from armored unit entry) and the discrimination of military incursions from common civilian activity.

The covert monitoring of military sites, for unusual activity and for triggering on significant events.

Urban Surveillance and Monitoring Missions and Tasks:

Security monitoring: search the exterior of strategic buildings for unauthorized incursion and unusual, possibly threatening activity (abandoned trucks or vehicles, vehicles left by persons not entering any nearby buildings, vehicles or persons maintaining surveillance of the building, etc.). This monitoring may also be for covert surveillance applications.

Activity understanding and discrimination: detecting, identifying and monitoring unusual civilian or military activity in urban settings (civilian gatherings; increased activities at airports or train stations; increased military movements; etc.)

3. Background

Offerors are encouraged to consult and cite any trade journal articles, research publications, conference reports, or program briefings relevant to VSAM technologies and goals. The offeror's attention is directed to the Proceedings of the annual DARPA IU Workshop as a source of current research and technology relevant to VSAM. See Section I, Part B for information on availability of these publications.

4. Enabling Technologies and Key Problems to be Addressed

The following discussions of possible VSAM research areas should be taken as suggestions only. Offerors are encouraged to propose any revolutionary and innovative video understanding technologies that may offer substantial benefit in future Battlefield and Urban VSAM missions.

Human, Animal, Vehicle, and Scene Motion Processing

(Including Modeling, Stabilization, Detection, Recognition,

and Reconstruction)

This category covers the processing of video streams from airborne, ground mobile, and ground stationary sensing platforms. A representative but not exhaustive list of possible technologies and tasks in this category is as follows:

The modeling, detection, and recognition of human motion and activity in video sequences. Motions may be at different scales, such as whole body motion (walking, running, crawling, jumping, falling) or partial body motions (waving, gesturing).

The modeling, detection, and recognition of animal motion, with the primary purpose of discriminating it from human motion in surveillance applications.

The modeling, detection, and recognition of vehicle motion and activity in video sequences for urban and battlefield surveillance. If vehicle recognition is proposed, it should incorporate motion integrally into the recognition process and offer very large potential gains in performance over single frame recognition. Simple tracking approaches are discouraged: efforts should seek to extend the state-of-the-art in terms of the richness of detail used by the algorithms as well as the performance of the algorithms

The recovery of the layout of space or shape of targets from multiple moving or stationary ground sensors. Such model creation should be designed to directly support VSAM tasks. For example, a moving ground sensor could construct a low-to-medium detail map of a village for the purposes of creating a world model to drive the platform's surveillance of the area. Note that the creation of detailed models from overhead imagery is the focus of the APGD Technical Area of this BAA.

Host motion processing or stabilization technology that may allow innovative development of VSAM capabilities using one or several moving (ground mobile and airborne) sensor host platforms.

Multiple sensor cooperative processing for any of the above methods.

Active or Task-driven Vision Using Single or Multiple Sensors

This topic generally refers to world-model-based, attention-driven, processing-based and/or task-based control and processing of video streams of imagery from multiple sensors.

World models could encompass things such as spatial models of urban and battlefield environments and temporal models of activities or common occurrences in sub-regions of the environment. "World-model-based", in general, means that VSAM systems will both: a) use the world model to drive video processing; and b) update the world model with processed video information. Offerors may design and interactively build such world models as may be necessary for their efforts, but the creation of spatial models is the primary focus of APGD section of this BAA.

"Attention-driven and processing-based" means that VSAM system reacts fairly rapidly to information derived from processed imagery streams: that specific detected events cause VSAM system reactions. "Reactions" may be events like: a) further sensing and processing with another sensor or a narrower field of view; b) the verification of other sub-activity in some functional area of the site; c) the gathering of specific imagery and the notification of the operator that inspection of such imagery is warranted.

"Task-based" means that VSAM system operation will vary with specific surveillance task.

"Multiple sensors" implies that the sensors cooperate to perform tasks, sharing sensing and computation in the creation and maintenance of an appropriate world model.

Activity modeling, Recognition and Understanding

"Activity modeling and understanding", in general, refers to the "medium scale" spatio-temporal activity of a single or few agents (humans or vehicles) as opposed to the small motion scales and the large world model scales listed above. For example, hand waving is a motion, while "opening a door" and "one man handing a box to another" are activities. One moving military vehicle could be understood with a motion model, but a higher level model may be needed to identify that it is digging a revetment or fording a stream.

The same general guidance applies to activity-based video processing as the "world-model driven" category above (i.e. attention-driven, processing-based and/or task-based control and processing of video streams of imagery from multiple sensors).

5. Technologies Not of Interest in This Technical Area

The following technology areas are not of interest in the VSAM Technology Area:

Development of automatic target detection or recognition technology that works on single frames of imagery. Such technology may be used as part of a more complex video surveillance system, but is not of primary interest here. These techniques are of interest in Technical Area 4 of this BAA.

Single frame image processing, image compression techniques, feature extraction technology, or computational speed-ups of existing techniques. Such technology may be used as part of a more complex video surveillance system, but is not of primary interest here.

6. Experimental Support

The imagery described in Section VI (Additional Resources & Experimental Data) is not relevant to VSAM. The government is not providing data for use in preparing VSAM proposals.

The following general guidance is offered with respect to experimental support (data, scenarios, laboratories, experiments, field experiments, evaluations, etc.) necessary for the VSAM effort:

Because cameras, digitization, computation, special image processors and storage equipment have become relatively inexpensive, it is anticipated that all research sites would be able, in some way, to gather data and perform experiments at their own location. It is encouraged that such on-site efforts be creatively designed to, at some level, be relevant to urban and battlefield VSAM missions.

The IFD developer should propose sufficient scope and effort for the support of FRE data collections, experiments and demonstrations at sites that are suitable for urban and battlefield VSAM scenarios.

DARPA will seek alliances with military and government labs to help focus VSAM efforts on relevant scenarios, foster technology transfer, and facilitate usage of government assets (equipment, test ranges, sensors) in VSAM development, demonstrations, and evaluations.

The FRE offerors are encouraged, wherever possible, to be strongly self-sufficient in terms of experimental support needed for research.

All offerors are encouraged to submit proposals with results from data collected at their own sites that is illustrative of the technology development proposed.

Both IFD and FRE offerors should include plans for pooling collected data and making it available to other researchers in order to foster broader VSAM technology development and common data sets for development and evaluation.

7. Guidance Specific to Categories

Category A: Integrated Feasibility Demonstration (IFD) for Video Surveillance and Monitoring (VSAM)

Goals: The goals of the VSAM IFD are:

to integrate, demonstrate and evaluate the prototype VSAM system for urban and battlefield applications, incorporating technology from any appropriate technology source, including other VSAM-sponsored FRE researchers,

to create and operate a VSAM testbed (for data collection, experiments and quasi-operational evaluations) and provide an information clearinghouse to support the general goals of the VSAM effort,

to perform innovative research by the IFD developer, and

to support, where feasible, any FRE effort that needs access to the IFD testbed for data collections, experiments or evaluations.

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document:

In the "Visionary Section" of the IFD proposal, the offeror should clearly define future operational scenarios for urban and battlefield VSAM systems and their embedded technologies. The following should also be defined: the technologies to be developed, the system architecture for the technologies, technology assessment, technology evolutions, experimentation plans, plans for periodic demonstrations during the project, and technology transfer plans.

The proposal should describe the creation of a VSAM testbed consisting of one host control unit, at least 2 sensor platforms, and all necessary software, processing hardware, sensors and sensor control hardware. The sensor platforms must contain processors sufficient for video rate computation and must allow sensor control reasonable for video surveillance (pan, tilt, zoom, focus, tracking). The platforms may be primarily stationary, but should have some capability to be mounted on moving ground vehicles to support host-motion based VSAM data collections and experiments. The platforms must allow for day and night operation and shall include surveillance-grade thermal or low-light cameras. The software will be based on a generalized VSAM architecture (developed by the IFD contractor) designed to allow host control unit monitoring multiple platform task and computation sharing. The testbed shall be designed to allow researchers at other geographic sites in the US to conduct testbed experiments via remote, internet-based access. The testbed shall be developed incrementally, allowing at least one VSAM platform operation as early as possible after program award. The design of the host control unit and VSAM platforms shall support periodic and temporary re-deployment to different urban and military areas to support data collections and field experiments.

The proposal should describe the use of the testbed for innovative experiments and evaluations of VSAM technology components. The components may be developed by the VSAM IFD developer or the FRE developers. The experiments may be conducted with the technology developers on-site or via remote, internet-based access. The experiments will occur in suitable urban or military environments as required for the IFD and FRE goals.

The proposal should describe the development and integration of unique, innovative technology by the IFD contractor, selected to both extend the state-of-the-art and also to enhance the infrastructure of the VSAM test bed as well as methods to be used to facilitate the integration of technology from FRE contractors.

The proposal should describe the creation of a general VSAM knowledge base and clearinghouse, containing information and data about future missions, concepts of operation, high level technology needs, and detailed technology requirements for VSAM technologies as well as technical descriptions, technical documentation, development schedules and software for existing and emerging VSAM technology components. As part of this effort, the VSAM IFD will coordinate, conduct and document two annual VSAM workshops attended by IFD personnel, FRE developers, government sponsors and potential users to: a) define VSAM-wide missions, concepts, and technology needs; b) review VSAM-wide data and experiment needs along with data collection and experiment plans; c) review ongoing IFD and FRE development; d) host the annual IFD demonstrations; and e) host such annual FRE demonstrations that the FRE developers choose to bring to the workshops.

The proposal should describe the conduct of frequent data collections and distributions designed to facilitate technology development, experimentation, and evaluation. The collections will occur in suitable urban or military environments as required for the IFD and FRE goals.

The proposal should describe how fundamental research, broad FRE community support, and integrated prototype demonstrations will be carried out by the offeror.

Category B: Focused Research Efforts (FREs) for Video Surveillance and Monitoring (VSAM)

Goals: The goals of the VSAM FREs are:

  1. to soundly design experimental research and development in robust and reliable video understanding techniques that may benefit VSAM applications (see "Enabling Technologies, Key Problems to be Addressed" above), and
  2. to cooperate with the IFD developer in advancing VSAM area goals and technology.

Guidance for Generating Proposal:

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document:

Innovative solutions to developing VSAM-related technology in a research setting are sought. Wherever possible, the FRE proposals should describe how all necessary sensors, data, controls and processing resources needed for the basic research are available, or will be acquired at the research site, from the beginning of the effort.

The FRE proposals shall also define how the research center will cooperate with the VSAM IFD in the creation of a general VSAM knowledge base and clearinghouse, containing information and data about future missions, concepts of operation, high-level technology needs, and detailed technology requirements for VSAM technologies.

The FRE proposals should describe what VSAM-wide data collections and experiments and evaluations that are necessary to achieve the research goals of their FRE development.

If necessary for their research goals, the FRE developer should describe what usage of the IFD testbed (generally described above) may be required.

8. Guidance for IFD and FRE Collaboration in the VSAM Program

FRE researchers are required to participate in semi-annual VSAM-wide meetings coordinated by the IFD developer.

FRE researchers are not required to contribute and integrate technology into the annual IFD demonstrations: they may choose to do independent development and demonstrations if this is more appropriate.

In consensus decisions, the IFD developer, the FRE researchers and the sponsor may decide that it is appropriate for an FRE component to be integrated into an IFD demonstration. After such decisions, the IFD developer and FRE researchers will coordinate plans to do so.

If feasible, FRE researchers are encouraged to be self-sufficient in terms of data collection and experiment capability. However, this should not limit the range of technical research performed.

If necessary to advance research goals, the FRE researchers are encouraged to work with the VSAM team at IFD-conducted workshops to define program-wide data collections and field experiments that support their research.

The FRE researchers are encouraged to nominate data collections, field experiments and evaluations that can only be achieved with the IFD test bed or broad VSAM participant cooperation. These collaborations should be well defined and effectively use available resources. The IFD developer, the FRE researchers and the sponsor will develop priorities and schedules to carry out such collections, experiments and evaluations.

D. Technical Area 4: Automatic Target Recognition - University Research Initiative (ATR-URI)

1. Objective

The objective of this Technical Area is to procure innovative FREs to support the ongoing DARPA-sponsored ATR-URI. Novel techniques for improving the performance of ATR applied to SAR imagery are the highest priority, but approaches for dealing with other image modalities are also desired.

Proposed research must address revolutionary ATR techniques for SAR, EO, IR, low-light television (LLTV), Video (both visible red/green/blue and black/white cameras), and LADAR imagery collected from airborne or ground-based surveillance and targeting platforms. Proposals must focus on very challenging ATR problems, have a significant emphasis on algorithm evaluation, and describe in detail the data available for research and evaluation. Proposers are encouraged to form alliances with industrial partners to both focus their research on relevant problems and facilitate the transfer of research techniques into applications development. The term "alliances" need not imply formal sub-contractual relationships: it may imply a non-contractual memorandum of understanding (MOU) that details cooperation in definition of URI research goals and schedules and the transfer of URI technology to other contractual development efforts undertaken by a commercial contractor.

Proposals must describe research that explicitly addresses very challenging ATR problems. It is appropriate to focus on a particularly challenging and unsolved area (for example, on targets 30% occluded in SAR imagery). A representative, but not exhaustive list of challenging target deployment and environmental conditions is as follows:

All research proposals must have strong, integral program plans for periodic and meaningful in-house evaluation of the proposed techniques. These in-house evaluations should follow commonly accepted industry and academic standards for ATR evaluations. Researchers must also make their algorithms and systems available for independent external evaluations during the program. DARPA will seek arrangements with government laboratories to conduct such evaluations.

DARPA is strongly interested in the research and development of systems that can be evaluated in terms of ATR effectiveness as opposed to sub-systems or component technology that cannot be evaluated in this manner. Proposers who focus on sub-system or component technology are challenged to develop evaluation plans that allow their developed technology to be evaluated in the context of overall ATR effectiveness. Proposals may be submitted only by US institutions of higher education (other than federal government) with degree-granting programs in science and/or engineering, or by consortia led by such institutions.

IFDs are NOT planned in this area. Approximately 5 to 10 FRE awards, none to exceed $200K/year, are anticipated in this category to maximize the variety of technical approaches being explored. Offerors should propose efforts consistent with the technical challenge of their effort and the equipment and staff-hour resources needed to address the challenge.

2. Background

Offerors are encouraged to consult and cite any trade journal articles, research publications, conference reports, or program briefings relevant to their proposed research.

3. Experimental Support

All research proposals must identify sources of data to support the specified research and evaluation. Wherever possible, the proposer should have possession of significant amounts of the data at the time of proposal submission. Offerors are encouraged to use data already available at their site or data provided by an industrial partner or government laboratory to support their research.

Wherever feasible, proposers are encouraged to purchase, lease, or otherwise arrange for the use of commercially-available or government-available sensors at their own sites or on available military sites to collect data needed for their research. They are encouraged to use terrain and targets that are technically analogous to real-world, ATR problems which are supportive of their particular research goals.

DARPA is in the process of seeking public release of the following to support university-based research. This data will be made available to contractors as approval is obtained.

-1 meter and 1 foot amplitude SAR imagery of rural scenes, villages and target arrays collected with the Advanced Detection Technology Sensor (ADTS) of MIT Lincoln Laboratory.

- IR, LADAR and Video RGB data of outdoor scenes and tactical ground targets collected under the DARPA Unmanned Ground Vehicle (UGV) Program.

4. Guidance for Generating ATR-URI Proposals

The following guidance supplements that provided in the "Introduction" and "Proposal Preparation" sections of this document:

The payoff of the proposed research effort on the advancement of ATR and how it addresses challenging ATR problems must be clearly and strongly presented.

The proposal must outline in, an orderly manner, the research to be carried out to achieve the research goals and the evaluations to be used to measure the degree of satisfaction of those goals.

Datasets needed for the research and evaluations must be clearly described and available.

Roles and responsibilities in university collaborations with industrial partners or government labs must be clear and achievable.

III. GENERAL INFORMATION

Independent organizations and teams are encouraged to submit proposals in any or all areas. However, offerors must be willing to cooperate and exchange software, data and other information in an integrated program with other contractors. Coordination with a prime contractor or integrator, chosen by DARPA, if appropriate. A statement of cooperation must be included in the proposal.

Software developed under the IU program will be shared with all IUBA participants to the maximum extent possible. Each proposal must discuss how proposed results will be sharable within the IU community and what use such results will be to others. Any restrictions on software or data developed under DARPA proposals that would affect this practice must be clearly identified in the proposal. This includes use of proprietary software support and development tools. All contractors may be required to enter into associate contractor relationships subsequent to contract award in order to better facilitate technology integration and transition through IFDs or other mechanisms.

A. Abstract Preparation and Delivery

Offerors, either individual or teamed, are strongly encouraged to submit an abstract of their proposed work. It is not a requirement for submission or selection of a proposal. Any offeror whose abstract is found to be consistent with the intent of this BAA will be invited, by surface mailed letter, to submit a full technical and cost proposal. Such an invitation does not assure subsequent contract award. Regardless of the recommendation, the decision to submit or not submit a proposal is the responsibility of the party submitting the abstract. Offerors will be notified of evaluation results upon completion of proposal evaluation.

Proposals and proposal abstracts may NOT be submitted by fax or e-mail. ABSTRACTS AND PROPOSALS SENT BY FAX OR E-MAIL WILL BE DISREGARDED. Letters of abstract receipt will not be sent to offerors.

Abstracts shall be submitted in the following format:

8.5 x 11 inches, 1.5 or double spaced

margins not less than one inch

numbered pages

at least 12 point type;

one sided printing

limit of 8 pages, including cover page

Cover Page - clearly labeled "PROPOSAL ABSTRACT," and must have the same format and information as the proposal cover page, as indicated in Part B, Volume I (A) of this section.

The summary (seven pages maximum) should include:

  1. a visionary system description that supports the goals of the BAA, and innovative ideas proposed,
  2. a comparison of those innovative ideas with current approaches,
  3. the expected impact of the research if successful,
  4. the process and metrics recommended for measuring the competence of the embedded IU technologies,
  5. brief description of a demonstration program that supports this vision,
  6. the anticipated project deliverables, and
  7. estimated cost per year.

An original and six (6) copies of each abstract referencing BAA 96-14 shall be

submitted to:

BAA 96-14
Defense Advanced Research Projects Agency
Information Systems Office/Attn: Thomas M. Strat
3701 North Fairfax Drive
Arlington, VA 22203-1714

Abstracts are due by 4:00 PM, EDT on Friday, 24 May 1996.

B. Proposal Preparation and Delivery

Proposals shall consist of two separately bound volumes (avoid using 3 ring binders). Volume I shall provide the technical proposal and management approach and Volume II shall address cost.

Proposals shall contain a Table of Contents, in both Volumes I and II.

The proposals shall be prepared in the following format:

8.5 x 11 inches, 1.5 or double spaced

margins not less than one inch

at least 12 point type

numbered pages

one sided printing

Page Limitations: Volume I of the proposal shall consist of the cover page, plus not more than 41 additional pages (excluding sections Q and R); foldouts shall be counted as a single page. Volume II is limited to 20 pages (excluding the cover page). Limitations within sections are indicated in the individual descriptions.

Note: Proposals with less than the maximum number of allowed pages will not be penalized. Proposals exceeding the page limit will not be reviewed. Offerors are encouraged to submit concise, but descriptive, proposals.

1. VOLUME I: Technical Proposal

Volume I of the proposal shall include the following sections, each starting on a new page:

A. Cover Page: This must include:

  1. the BAA number,
  2. title of proposal,
  3. identity of prime offeror,
  4. complete list of subcontractors, if applicable,
  5. technical contact (name, address, phone/fax, electronic mail address),
  6. administrative contact (name, address, phone/fax, electronic mail address),
  7. technical area being proposed -- one of:

Image Exploitation

Automatic Population of Geospatial Databases

Video Surveillance and Monitoring

Automatic Target Recognition (Focused Research Efforts only),

  1. category -- one of:

Category A - Integrated Feasibility Demonstration or

Category B - Focused Research Effort,

  1. type of business (large business, small disadvantaged business, other small business, HBCU or MI, other educational, or other nonprofit),
  2. duration of effort (differentiate basic effort and options),
  3. proposed costs: the following table of cost data should be included on the cover:
YEAR	*CY 1997	CY 1998		CY 1999		CY 2000		CY 2001 
MONTHS	Jan-Dec		Jan-Dec		Jan-Dec		Jan-Dec		Jan-Dec 
BASE	$ XXX		$ XXX							$Total Base 
OPTIONS					$ XXX		$ XXX		$ XXX	$Total Options 
TOTAL Cost									$ XXX 
*CY=Calendar Year 

B. Executive Summary: (Limit: 3 pages) The summary should include: (1) a visionary system description that supports the goals of the BAA, (2) innovative ideas proposed, (3) the expected impact of the research if successful, (4) description of capabilities to be demonstrated, and (5) the major project deliverables.

C. Innovative Claims: (Limit: 1 page) This summary of innovative claims must identify any technical ideas to be pursued and their expected impact on the state-of-the-art. This page should succinctly describe the unique proposed contribution.

D. Vision: (Limit: 3 pages) This must describe a hypothetical, yet relevant, advanced battlefield awareness concept. Explain how the proposed technology development would contribute to improved battlefield awareness, and how it could be incorporated into current or hypothetical military or intelligence information systems. What minimum level of IU competence would be needed to realize the vision? What additional technology beyond the scope of the proposed effort would be necessary to realize the vision?

E. Deliverables and Products: (Limit: 4 pages) This section must enumerate the deliverables of the proposed effort, and the due date for each. The list must include those deliverables specifically required by this BAA, plus any additional deliverables that are offered.

F. Schedule and Milestones: (Limit: 1 page) A summary of the schedule and milestones for the proposed effort. Milestones must be specific and goal or performance oriented.

G. Proprietary Claims: (Limit: 1 page) Include here a summary of any proprietary claims to results, prototypes, or systems supporting and/or necessary for the use of the research, results, and/or prototype. Any claims made in other parts of the proposal (such as Sections B and E) that would impact the claims in this section must be cross referenced. If there are no proprietary claims this section shall consist of a statement to that effect.

H. Statement of Work (SOW): (Limit: 5 pages) This section must detail the scope, background, objective and approach of the proposed effort and describe the content and timing of specific tasks to be performed and specific utilization of subcontractors. Include here a detailed listing of the technical tasks/subtasks organized by contract year. Also identify which personnel and subcontractors (if any) will be involved. The SOW should contain optional tasks to support technology transition as called for in this document, under description of Approach.

I. Technical Rationale: (Limit: 8 pages) The technical rationale section must include technical arguments to substantiate claims made in Section B, a technical approach description consistent with Sections C and D, and a comparison with other ongoing research indicating both advantages and disadvantages of the proposed effort/approach. Include a discussion of design decisions made. The technical approach must include a Demonstration Plan.

J. Evaluation Plan: (Limit: 4 pages) This section describes the critical experiments to be performed, the data to be used, the evaluation metrics to be applied, and the software instrumentation and test plans to facilitate repeatable experimentation. This section must include a description of the metrics (especially for enabling technology efforts) that could/will be used for evaluating the impact of the proposed effort, and the performance goal needed to achieve the vision described in Section D.

K. Management Plan: (Limit: 2 pages) This section describes the overall approach to management of this effort, including a very brief discussion of the organization, use of personnel, project/function/subcontractor relationships, government research and facility interface, and planning, scheduling and control practices.

L. Technology Transition Plan: (Limit: 2 pages) This section should contain a clear description of how results will be made sharable throughout the IU program and what use these results might be to other groups. In addition, this section should address specific innovative approaches the offeror will take to facilitate technology transition (e.g., send graduate students to a partner company for "a summer with industry," conduct a summer school, institutionalize newly developed techniques by implementing them as an extension to the IUE, etc.). The technology transition plan should identify the potential recipients and describe the overall approach to delivering results, products and technology.

M. Facilities: (Limit: 1 page) Include here a description of the facilities that would be used for the proposed effort.

N. Security Plan: (Limit: 2 pages) State what level of security is required to carry out the proposed effort, if any. If access to classified information is required, provide information specific to secure facilities, including storage and disposal of classified materials, security personnel (officer) and procedures to be employed. Also include the clearances in place and those that would be required to carry out the project. Discuss limitations foreseen. ALL PROPOSALS MUST BE UNCLASSIFIED.

O. Experience: (Limit: 3 pages) This section describes relevant capabilities, accomplishments, and work in these or closely related areas along with the qualifications of proposed subcontractors.

P. Key Personnel: (Limit: 1 page) Include a listing of key personnel along with the amount of effort to be expended by each person during each calendar year. If multiple proposals are being submitted in response to this BAA, indicate how they will be staffed if multiple awards are made.

Q. Qualifications: (Limit: 1 page per key person) This section contains a concise summary of the qualifications of listed key personnel along with other major sources of support for them. (This section is not included in the page limit.)

R. Bibliography: Include here a bibliography of relevant technical papers and research notes which support the technical ideas in this proposal. (This section is not included in the page limit.)

2. VOLUME II: Cost Proposal

In general, the cost proposal should provide for a phased program over the duration of the project, supported by detailed breakdowns. Volume II of the proposal shall consist of a (A) Budget Cover Page and not more than 20 additional pages of detail, B) Budget Summary, part 1 and 2, and C) Budget Details. Details of any cost sharing to be undertaken by the offeror should also be included in the cost section. An SF1411 is required for the submission of your proposal.

  1. Cover Page: This must include the words "Cost Proposal" and shall otherwise be identical to the Volume I cover page as described in Volume I, Section A.
  1. Budget Summary:

Part 1 (one page): Detailed breakdown for all costs by calendar year:

  1. labor hours by labor category / tasks and subtasks; optional tasks/subtasks must be listed individually.
  2. personnel (name or designation, rate, and percent time on project, and other miscellaneous expenses, such as benefits/overhead rate
  3. proposed contractor-acquired equipment (computer hardware for proposed research projects) should be specifically itemized with costs or estimated costs. An explanation of any estimating factors, including their derivation and application, shall be provided. Please include under Budget Details a brief description of the offeror's procurement method to be used.
  4. travel
  5. other direct / indirect costs
  6. materials

Part 2 (one page): Cost breakdown by task / sub-task using the same task numbers as in Technical Proposal SOW, Volume I, section H. Options must be costed individually.

C. Budget Details: Include any other relevant details that support section A above.

Protection of Information: It is the policy of DARPA to treat all proposals as competitive information and to disclose the contents only for the purposes of evaluation. The Government may use selected support contractor personnel to assist in administrative functions only.

C. Proposal Delivery

An original and eight (8) copies of the Technical and Cost proposals must be submitted to:

BAA 96-14
DARPA/ISO
Attn: Thomas M. Strat
3701 North Fairfax Drive
Arlington, Virginia 22203-1714

Proposals are due by 4:00 PM, EDT on Tuesday, 23 July 1996

IV. EVALUATION CRITERIA

Proposals and abstracts will be selected through a technical/scientific/business decision process with technical and scientific considerations being most important. Evaluations will be performed using the following criteria listed in descending order of relative importance:

Criteria 1- 4 are listed in order of priority.

  1. Quality and Technical Merit:
  1. Capabilities and Experience
  1. Relevance to Image Understanding for Battlefield Awareness
  1. Approach to Technology Transfer

Realism of cost proposal will be evaluated separately.

  1. Cost realism and value of proposal to government

V. TECHNICAL AND ADMINISTRATIVE INFORMATION

  1. Technical or contractual questions shall be emailed to <[email protected]> or faxed to (703) 516-6065, attn: BAA 96-14. Refer to FAQ file for answer.
  2. Access to Proposer Information Pamphlet (PIP) and FAQ file: DARPA provides this Proposer Information Pamphlet (PIP), as well as a BAA 96-14 FAQ file, on WWW, and by automated e-mail response system.
  3. Web Site: URL: http://www.darpa.mil/baa/. BAA96-14 CBD Announcement, PIP and FAQ file appear under Information Systems Office.
  4. Automated e-mail responder: Note: The system is CAP SENSITIVE--use lower case only. For auto-responder access to ALL available files on BAA 96-14 (will include PIP, FAQ and any other support files that can be transmitted in text) send an email message to <[email protected]> with NO SUBJECT. Within the body of message enter: "info baa 96-14" (without quote marks). You will receive a listing of all available files and instructions on how to access those files via email auto-response.
  5. Facsimile: If the offeror does not have access to e-mail or WWW, a request for information may be faxed to 703-516-6065 attn: IUBA BAA 96-14. Clearly indicate the items requested, and include necessary information (phone number, fax number, surface mail address) to insure proper delivery.
  6. Surface mail: If the offeror does not have access to e-mail, WWW or a FAX machine, written requests may be sent to: BAA 96-14 INFORMATION, 4301 N. Fairfax Drive, Suite 700, Arlington, VA 22203. Clearly indicate the items requested, and include necessary information (phone number, fax number, surface mail address) to insure proper delivery. Be aware that surface mail will require more response time than other methods.

VI. ADDITIONAL RESOURCES: Experimental Data

Imagery data for proposal preparation and for initial experimental support will be available from publication date of BAA96-14 until 15 July 1996. Use of this data is encouraged but not required. It is being provided by the government as an aid to proposal preparation. Upon request from a proposer, only one tape per organization will be provided. This is a summary of the data:

Included on the data tape are the following:

Imagery Exploitation: SAR data and optical imagery of Stockbridge, NY

Automated Population of Geospatial Databases: Optical imagery of Ft. Hood.

Both are included on one 8 mm Digital Tape.

A description of the Ft Hood dataset and associated metadata is available on the WWW at URL: http://www.mbvlab.wpafb.af.mil/public/fthddata/

All requests should clearly indicate the accurate name, organization, surface mail address with zip, and phone number. Incomplete information will delay or prevent delivery.

Email requests are preferred. Proposer should send a clear request for "BAA 96-14 Experimental Imagery Data Tape" to <[email protected]>. If the offeror does not have access to e-mail, fax request for Experimental Imagery Data Tape to IUBA BAA 96-14 at (703) 516-6065, or surface mail written request to: IUBA BAA 96-14, 4301 N. Fairfax Drive, Suite 700, Arlington, VA 22203.

VII. OTHER IMPORTANT ADMINISTRATIVE INFORMATION

DARPA intends to use E-mail for some correspondence regarding BAA 96-14; however, proposals and proposal abstracts may NOT be submitted by fax or e-mail. ABSTRACTS AND PROPOSALS SENT BY FAX OR E-MAIL WILL BE DISREGARDED. The Government reserves the right to select for award all, some or none of the proposals received in response to this announcement. All responsible sources may submit a proposal which shall be considered by the Agency. Historically Black Colleges and Universities (HBCU) and Minority Institutions (MI) are encouraged to submit proposals and join others in submitting proposals, however, no portion of this BAA will be set aside for HBCU and MI participation due to the impracticality of reserving discrete or severable areas of technology for exclusive competition among these entities.

This PIP, along with the Commerce Business Daily (CBD) announcement, constitutes a Broad Agency Announcement (BAA) as contemplated in FAR 6.102 (d)(2)(i). Prospective offerors MUST also refer to this PIP before submitting an abstract or a proposal. DARPA anticipates that initial contractor selections will be made during the first quarter of Fiscal Year 1997.