Discovery Challenge Thrusts (DCTs) This section outlines cross-cutting multi-disciplinary topics that support the AFOSR’s Discovery Challenge Thrusts (DCTs). Research efforts will consist of interdisciplinary teams of researchers with the skills needed to address the relevant research challenges necessary to meet the program goals. Proposers are highly encouraged to confer with the appropriate AFOSR program manager. White Papers briefly summarizing your ideas and why they are different from what others are doing are highly encouraged, but not required. Coordination with the Air Force Research Laboratory is also encouraged but not required.
Air Force program managers are listed by Sub areas below.
Description: The Air Force Office of Scientific Research is seeking basic research proposals to conceive adaptive multi-modal EO-RF sensor concepts in a ‘performance-driven’ context that addresses the challenging problems of detecting, tracking, and identifying targets in highly cluttered, dynamic scenes. ‘Performance-driven’ requires that the development of novel adaptive multi-modal sensing hardware concepts be closely coupled with concurrent developments in novel physics-based modeling and simulation of target scene phenomenology, environmental interactions, and breakthroughs in data processing and exploitation. An integrated approach allows for assessing the utility of combining different sensing modalities, utilizing associated novel fused-data processing schemes for the target and background scenes of interest. It is expected that each research effort will consist of an interdisciplinary team having the appropriate skills needed to address all of the relevant program research challenges.
Background: The premise of this research is that developing adaptive multi-modal sensors able to capture multiple electromagnetic observables (intensity, wavelength, polarization, and/or phase) in a time-resolved, ‘staring’ imaging format will provide dramatically enhanced detection and identification capability for extremely challenging military problems involving low contrast targets over broad areas in a highly dynamic scene. Battlefield sensing requirements include finding and tracking individuals of interest in populated urban areas, detecting activity and materials indicative of IED placement, and detecting and identifying threatening space objects at long ranges. Historically, military target recognition involved conventional military objects exhibiting unique spatial and spectral signatures that were generally isolated from densely populated areas. However, today’s target recognition problems include discriminating a multitude of complex objects deeply embedded in urban areas, day and night, where the most common urban objects can have tactical significance, and achieving high detection probability is critical to mission success. Current-generation remote sensing methods (e.g., broadband FLIR) are limited in their ability to search and detect camouflaged targets in deeply-hidden or highly-cluttered backgrounds. Proven approaches for enhancing deeply-hidden, high-clutter target recognition includes utilizing multi- to hyper-spectral exploitation to improve signal-to-clutter ratio, and fusing multi-modal/multi-discriminant data, such as FLIR with SAR, to significantly reduce the amount of processing required for target classification, while simultaneously increasing target ID confidence.
However, limitations facing state-of-the-art multi- and hyper-spectral imagers include their ‘step-stare’ mode of operation (vs. desired staring mode) with revisit times that compromise detection of rapid moving targets, and their fixed-multi/hyper-band construct that can result in a tremendous amount of unimportant data for exploitation. Also, today’s airborne hyper-spectral sensors are massive, typically 4-5X that of typical FLIR sensor units employed on tactical aircraft and weapons platforms, and they also require greater sensitivity than typical FLIR sensors to overcome the reduced photon count in narrow wavelength bands. Challenges confronting fusion of multi-discriminant data from single-mode detectors include handling translational registration errors, and a lack of robust, efficient feature extraction and correlation capabilities. To avoid the problems of unnecessary or unproductive sensor use and computations, it would be desirable to ‘intelligently’ select ‘on-the-fly’ an optimum subset of sensors and sensor settings that are most decision-relevant. While this will be very difficult, requiring breakthroughs in many sensing technology fronts, emerging innovations in semiconductor materials, device structures, and information sciences offer many interesting opportunities. A ‘home-run’ approach of interest is to innovate and develop a tunable multi-mode, vertically-integrated (common sensor package), large-format staring focal plane array to accommodate the dynamic sensing requirements dictated by the dynamic target scene. This would involve actively controlling sensor modes and settings to optimize information gathering in a knowledge-based manner with an identifiable selection criterion.
Basic Research Objectives: Program focus is on modeling and simulation of novel concepts for high-performance tunable multi-modal EO-RF focal plane arrays. This includes innovative physical device concepts and prediction of single- and fused-mode detector output signals, in coordination with first-order benefits analysis modeling of downstream data exploitation. Novel multi-modal detector designs should be guided by consideration of how they can optimally exploit the phenomenology of multi-modal target scene signatures; and of how multi-mode data streams can be fused and interpreted in novel and beneficial ways. For example, fused spectral-polarimetric signatures provide information on target material composition, surface characteristics, and 3-D shape simultaneously from a single sensor snapshot, where information in the spectral dependence on polarization state may not be evident from separated polarization and spectral data. To exploit these and other multi-mode opportunities, a closely coordinated multi-discipline research team, expert in detector device design, data fusion, and image processing and exploitation will be needed. While the primary focus of basic research is on innovative integrated multi-modal EO-RF detector device concepts, supportive analysis and understanding of downstream data exploitation utility will be essential. Sensing modalities of interest include spatial, spectral, polarimetric, radiometric, and temporal; wavelengths of interest span UV (0.2um) to RF (mm). The envisioned multi-modal device design should build from extensive developments in both passive and active sensing, but specifically address the basic research aspect of multi-modal integration into a common sensor package (e.g., detector array). The ultimate vision would be a starring sensor development approach that optimizes the collection of phenomena to support detection, tracking, and identification functionality. The sensor would capture, at the pixel level, the right combination of the pixel intensity spectrum, polarization state, time evolution (at high enough bandwidth to capture active ranging and vibration information), and possibly phase (field vs. intensity), and work cooperatively with other sensors to perform such functions. This sensor would be accompanied by a high fidelity model to confidently predict its performance as a function of sensor configuration and target and background characteristics. It is expected that proposals will describe cutting-edge efforts on basic scientific problems.
Program Scope: Single awards will typically be $250-300K per year, for 3 years. It is expected that each research effort will consist of an interdisciplinary team with the skills needed to address all of the relevant research challenges necessary to meet the program goals. Multi-university teaming is encouraged.
Dr. Douglas Cochran/AFOSR/RSL (703) 696-7796
FAX (703) 696-7360
2. Robust Decision Making
Description: The need for mixed human-machine decision making appears at all levels of Air Force operations and pervades every stage of Air Force missions. However, new theoretical and empirical guidance is needed to prescribe maximally effective mixtures of human and machine decision making in environments that are becoming increasingly complex and demanding as a result of the high uncertainty, complexity, time urgency, and rapidly changing nature of military missions. Massive amounts of relevant data are now available from powerful sensing systems to inform these decisions; however, the task of quickly extracting knowledge to guide human actions from an overwhelming flow of information is daunting. Basic research is needed to produce cognitive systems that are capable of communicating with humans in a natural manner that builds trust, are proficient at condensing intensive streams of sensory data into useful conceptual information in an efficient, real-time manner, and are competent at making rapid, adaptive, and robust prescriptions for prediction, inference, decision, and planning. New computational and mathematical principles of cognition are needed to form a symbiosis between human and machine systems, which coordinates and allocates responsibility between these entities in an optimal collaborative manner, achieving comprehensive situation awareness and anticipatory command and control.
Basic Research Objectives: In the area of a) data collection, processing, and exploitation technologies, there is a need for
(a.1) attention systems for optimally allocating sensor resources depending on current state of knowledge,
(a.2) reasoning systems for fusing information and building actionable knowledge out of raw sensory data,
(a.3) inference systems for real time accumulation of evidence from conflicting sources of information for recognition and identification.
In the area of b) command and control technologies, there is a need for
(b.1) prediction systems for anticipating future behavior of adversarial agents based on past experience and current conditions,
(b.2) rapid decision systems with flexible mixtures of man and machine responsibilities for reactive decision making under high time pressure,
(b.3) robust strategic planning systems designed to allow for sudden changes in mission objectives, unexpected changes in environment, and possible irrational actions by adversaries.
In the area of c) situation awareness technologies, there is a need for a human-system interface that
(c.1) faithfully simulates the content of a human operator’s working memory buffer and its update thus modeling the operator’s dynamic awareness of inputs, constraints, goals, and problems,
(c.2) optimizes information delivery, routing, refreshing, retrieval, and clearance to/from the human operator’s awareness while utilizing the latter’s long- term store for expert knowledge, memory and skills for robust decision making,
(c.3) achieves symbiosis between human and machine systems in delegating and coordinating responsibilities for command and control decisions.
In sum, new empirical and theoretical research is needed that provides a deeper understanding of the cognitive requirements for command and control by a decision maker with enhanced capability for situation awareness, allows for greater degree of uncertainty in terms of reasoning systems, produces greater robustness and adaptability in planning algorithms in dealing with unexpected interruptions and rapidly changing objectives, generates greater flexibility in terms of assumptions about adversarial agents, and gives clearer guidance for dealing with the complexities encountered in network-centric decision tasks. Projects that bridge the conceptual gaps between state-of-the-art statistical/machine learning algorithms or AI systems and human cognition and performance are particularly welcomed.
Program Scope: Typical awards could be $100-200K/year. It is expected that each research effort will consist of an interdisciplinary team formed from some combination of cognitive, computer, engineering, and mathematical/statistical scientists having the appropriate skills needed to forge new breakthroughs and make significant fundamental progress in this area.
Dr. Jun Zhang/AFOSR/RSL (703) 696-8421
FAX 703 696-7360
3. Turbulence Control and Implications
Description & Background: Under the AFOSR Discovery Challenge Thrust: Turbulence Control and Implications, AFOSR is pleased to solicit basic research proposals addressing the exploration, characterization, and modeling of fundamental processes in transitional and turbulent flows including, but not limited to, flow regimes characterized by either low Reynolds number or compressibility. Specific topics of interest for this BAA include the following.
Basic Research Objectives:
Effective actuation in flowfields relevant to AF systems that exploits flow physics (flowfield bifurcations, instabilities, etc.) and responds to a dynamic environment, with the following qualities:
• Robust, scalable actuation with adjustable authority as required by flow conditions. Both passive and active approaches may be considered.
• Characterization of the effectiveness of flow control methods, considering the influence of actuation rate and phase with respect to flow structures for tailored amplification or attenuation of disturbances.
• Development of robust, reliable sensors for flow control. Desired sensors should be adaptive, embeddable in the system, and possibly self-powered. Sensors should measure surface shear stress, pressure, or another physical quantity useful for inferring the flow state. Ideal sensors will be sensitive to very-low-amplitude disturbances with high spatial- and temporal-resolution and signal-to-noise ratio. Integration into limited-size wind-tunnel and flight experiments also must be considered.
High-fidelity models of transitional and turbulent flows incorporating flow control: Models should enable characterization and reliable prediction of physical phenomena associated with flow control, including transient and dynamic processes. Additionally, the models developed under this thrust should enable the development of reduced-order models for complete potentially-fielded flow control methods to facilitate design requirements and optimization without compromising other mission aspects.
Research areas of interest under this topic include, but are not limited to, the following:
• Incorporation of multi-disciplinary analysis (e.g., aerodynamics, structures, materials, controls, sensing and actuation) including transfer of the proper physical quantities between sub-models for various disciplines.
• Integration of experimental, numerical and theoretical analyses.
• Development of advanced diagnostics required for characterization of the fundamental phenomena associated with flow control methodologies and for validation of numerical simulation tools.
Ideally, basic research efforts supported under this BAA will have relevance to a wide variety of potential applications. Air Force interest in the research solicited under this portion of the BAA includes, but is not limited to, potential application to the following flows:
• Compressible flow at high-subsonic, transonic or low-supersonic conditions for flight vehicles intended to efficiently operate across several speed regimes.
• Low-Reynolds number unsteady flows encountered by agile micro air vehicles.
• Transonic compressible flow over aero-optic turrets and cavities.
• Unsteady flows generated by high-lift systems, propulsion systems and landing gear responsible for significant acoustic emissions.
Program Scope: Both single- and multi-investigator proposals will be considered, with typical awards in the range of $100k-$300k.
Dr. John Schmisseur/AFOSR/RSA (703) 696-6962
FAX (703) 696-8451
4. Space Situational Awareness
Description: The Air Force Office of Scientific Research is seeking basic research proposals to develop concepts and capabilities in the area of Space Situational Awareness (SSA). The goal is to detect, track, identify, and predict future capabilities, actions, and positions of all space objects at all altitudes with known accuracy and precision. This capability must include on-demand capacity for a highly-detailed characterization of individual space objects. SSA is more than the observation of the location and orbit of an object in space or the image of the object; it must include the ability to identify a satellite’s capabilities and predict future operations and performance limits with known confidence. Therefore, we must be able to detect and understand the configuration and orientation of the satellite, and to detect and quantify maneuvers through changes in orbital state, object signature or telemetry, or characteristics of exhaust products. Prediction of the precise location of satellites and limitations to satellite operations requires knowledge of the space environment in near-real time and an understanding of the impacts of the space environment on space systems. Understanding of the physics of the environment is also required for accurate space environment forecast models.
Background: The challenge of SSA is to rapidly and accurately locate and comprehensively characterize every object in space with known confidence and in near real time, including its orbital parameters, physical state, purpose, and capabilities, to anticipate future actions based on real-time estimates of changes in state using all sources of possible information, and to appropriately and rapidly provide actionable, useful information. Predictive SSA helps to ensure the safe flight of satellites and to mitigate impacts from the space environment on operations. It provides the capability to identify, characterize and monitor all potential threats to friendly space assets and adversary space capabilities that pose a threat to friendly terrestrial forces and to make after-action assessments. SSA is long-term, immense in scope, continual in maintenance, and demanding in detail and timeliness.
Our space surveillance models, tools and sensors today have significant capabilities but are not adequate for the problems of the future. Space search and track requires observations over several orbits and may take from days to months for the identification of small and poorly resolved objects. In addition, data are limited by collection methods to specific orbital planes and local times for space-based observations and to specific locations for ground-based observations. Observations of small and distant satellites are especially problematic, as is the discrimination of these objects from space debris.
Knowledge of the space environment is an integral part of SSA. This knowledge is based on theoretical studies of a sparse data set of ground- and space-based remotely sensed data and in situ observations. Each observational modality has fundamental limits. Current models provide some capability of "nowcasting," but are limited by deficiencies in the physical understanding of the solar-terrestrial system. Much of the current forecasting capability is based on statistical or climatological models.
Basic Research Objectives: Successful proposals will propose research that addresses the current needs for space situational awareness described above.
Priority will be given to proposals that address basic principles and fundamental limits of the following:
1. Non-imaging techniques leading to the identification and characterization of un-resolved space objects.
2. Innovative solutions to the inverse problems associated with characterization of non-resolved space objects.
3. Novel imaging or image processing methods to fundamentally decrease limitations on remote imaging of space objects
4. Predictive analyses of space objects that include characterizing, tracking and predicting the behavior of individual and groups of satellites using multi-source data.
5. The resolution of uncorrelated tracks and marginally detectable targets using sparse data.
6. The physical processes that control the formation and growth of ionospheric irregularities that impact communication, navigation and radar systems.
7. Phenomenology and basic physical processes leading to the understanding and forecasting of the neutral atmosphere and ionosphere.
Program Scope: The typical awards will be $150-250K per year for a three-year effort. Although it is expected that single investigator projects will be awarded, multidisciplinary team proposals will also be considered. Collaboration with researchers at the Air Force Research Laboratory is encouraged.
Dr. Kent Miller/AFOSR/RSE 703-696-8573
FAX 703 696-8481
5. Complex Networked Systems
Description: Air Force network systems today are faced with increasing demands on reliability and performance in many heterogeneous mission scenarios, network infrastructures, policies, and protocols. In order to address these challenges, we wish quantify the likelihood that critical information associated with specific mission needs will reach its destination with predictable latency rather than packets simply reaching their destination. Additionally, we would like to quantify the likelihood that a given network protocol or policy will support delivery of this information with a certain probability. We would also like to ensure that such policies will not lead to network instability due to lack of resources or introduce vulnerabilities in security. Finally, we would like to establish a comprehensive strategy to manage network content, protocol, policy, and network structure for highly heterogeneous and dynamic network conditions. Examples of such strategies may involve in distributed network coding, estimation, optimization, and routing techniques, that can recover and route information even if protocols fail or are interrupted. Additionally they include network analysis techniques that can detect or inference global network performance from many sparse distributed local measurements. These fundamental approaches to assessment and design of information exchange will then be used to improve overall network protocol performance, detection of and resilience to attack, scalability, routing performance, human network interaction, coding efficiency, resource utilization, throughput, latency, and reconfigureability as examples.
Basic Research Objectives: We thus wish to establish new methods to design and manage networks that assess and quantify performance at all levels and conditions of network operation. Areas of interest in ensuring predictable network performance include new methods for coding and quantization, new approaches for advanced rate distortion analysis, entropy, and error correction coding. We would also like a mathematical means of guaranteeing system performance in the context of dynamic network policies, human network interaction and decision-making, heterogeneous wired, wireless, and hybrid networks, and scalable numbers of users. We would like to explore methods of assessing the relative effect and interaction of different layered mission functions on a network including reconnaissance, distributed computation, platform positioning and control, and overall course of action prediction. At the networking level, areas of interest include new approaches to assessing the reliability of connections as a result of current and future protocol layering and buffering and cashing approaches, data retransmission, flooding, and latency. We would also like to develop new mathematical paradigms for quantifying centralized and decentralized routing performance and multiple access. In the area of physical transfer of data, we would like to understand new approaches to predictable space time coding, modulation, spectrum access, and physical routing mechanisms that are resilient to interference and attack.
Program Scope: Typical awards could be $125-250K per year for individual investigators. Multidisciplinary team proposals also are encouraged and will be considered on a case by case basis. Projects that include collaboration with scientists in the Air Force Research Laboratory are encouraged.