ML-RCP Opportunities

ML-RCP Opportunities

The AFRL publishes research topics through a Project Opportunity Announcement (POA). Faculty participants respond with white paper submissions. If a white paper is selected, the AFRL invites the faculty participant to submit additional information through a Request for Project Proposal (RFPP). AFRL independently reviews and selects all projects. 

Faculty participants receive complete instructions to assist them with the submission process. Selected proposals receive funding through a subaward agreement between the ML-RCP member institution and program facilitator The Ohio State University (Ohio State).

Project Opportunity Announcements (POAs)

Anyone may submit a white paper against a currently open Project Opportunity Announcement (POA). However, your institution must join the ML-RCP consortium to receive project funds. 

Below are the currently open AFRL ML-RCP POAs. 

Use the ML-RCP White Paper Template for your submission. Download the White Paper Template here.  Request the White Paper Template here. 

Submit your white paper using the submit button below. 

Currently Open POAs

There are currently no open POAs. Please contact mlrcp@afresearchlab.com with questions. 

CLOSED POAs

POA-24-RV-002: Local Intelligent Networks of Collaborative Satellites Laboratory

Closed: Review in process

Background 

The Local Intelligent Networked Collaborative Satellites (LINCS) Laboratory is a multi-agent robotics laboratory at the Space Vehicles Directorate. It is interested in developing and implementing novel algorithms for collaborative satellite autonomy. In this project, interns will work collaboratively in the LINCS laboratory in a team environment. Prospective interns should be independent and motivated to learn and gain new skills. They will work closely with senior researchers, contractors, and other interns to develop an understanding of the robotics platforms and their control mechanisms, build and design new robotics platforms and sensing capabilities to test, learn embedded systems programming and robotics skills. There is also the ability to practice and learn about novel ways of computer vision, control mechanisms in unique environments, and machine learning/reinforcement learning concepts.

POA-24-RX-001: Characterization and Performance Prediction of Advanced Composite Systems

Closed: Review in process

Background 

Advanced composite materials have provided game changing capability for both structures and propulsion systems in both aero and space systems. In particular, continuous fiber reinforced polymer and ceramic matrix systems are enabling the Dept. of the Air Force to operate systems longer, more efficiently and in more extreme environments than ever before. However, development of such systems remains complex, with significant time required for materials development, materials design and materials sustainment with excessive conservatism throughout the process. In addition, new step changes in performance will not be possible without multifunctional materials performance, where structural materials are able to also directly support other system requirements such as conductivity, energy storage, morphing abilities, etc. New approaches, frameworks and tool sets are required that will reduce overall system lifecycle cost from materials inception to system retirement. Additionally, future systems will require real time assessment of the materials systems health and ability to meet current mission requirements, which will only be possible with the advent of reduced order performance models. Finally, novel information rich materials characterization approaches are needed to enable automated three-dimensional materials quantification, high throughput materials experimentation, and rapid screening of key property sets all in application relevant environments. In many cases the application environments will require the ability to test and measure performance in extreme environments (e.g., high temperature, chemically corrosive, or radiatively intense).

POA-24-RX-002: Robot-Sensor Systems for Material Assessment in Access-Limited Spaces

Closed: Review in process

Objective
Advance the state-of-the-art in robotic material inspection, particularly in confined and complex spaces, through research related to autonomous inspection of complex structures, sensor integration with small or flexible robotic systems, and/or multi-agent teaming for navigation and inspection in complex environments.

POA-24-RX-003: Optimization of multi-objective outcomes in AM composites by scaling reinforcement density to FEA results

Closed: Review in process

Description
Demand continues to increase for composite materials that simultaneously satisfy multiple objectives, like structural components that maximize stiffness and fracture toughness while including embedded sensors. Recent advancements in local composition control through additive manufacturing (AM) have expanded the design space for composite materials by enabling variability in reinforcement and matrix constituents throughout a component’s geometry including the rare ability to fabricate compositionally graded structures. Despite these physical capabilities in AM, design constraints persist due to the current digital thread for planning print operations, which generally assumes uniformity in composition. Some AM technologies, such as polyjetting, are more conducive to fabricating tailored compositions due to a voxel-based print procedure but are not capable of printing with composite reinforcement as is of interest to this project. Alternatively, some recent advancements in extrusion-based AM methods have enabled local composition control with polymer-matrix composites. These novel capabilities have prompted an interest in developing toolsets for tailoring composition profiles in AM composite material structures for increased functionality. A step toward comprehensive composition optimization is proposed in this study by scaling composition to Von Mises stress FEA results.

POA-24-RX-004: Design for Manufacturing using Machine Learning

Closed: Review in process

Description
Recent advances in machine learning (ML) have shown great promise for improving automated design techniques such as topology optimization (TO), usually by demonstrating improved computational efficiency or superior mechanical performance. However, such research often focuses on design novelty and performance, while treating manufacturability as a secondary rather than coequal consideration. This helps perpetuate the proverbial wall between mechanical designers and manufacturing engineers. The present topic seeks to understand how ML-guided design optimization can best incorporate manufacturability. For example, how manufacturability constraints can be represented in an ML-friendly manner remains an important question. Additionally, manufacturability filtering tends to increase the chances of local optimality, and developing an understanding of how well ML frameworks operate under these conditions is key to their further adoption in practice. Proposed research can focus on manufacturability for multiple manufacturing processes or focus on a single process in greater detail. 

POA-24-RD-001: Ultrashort pulse laser atmospheric propagation 

Closed: Review in process

Background 

Ultrashort pulse lasers with pulse durations from 10s of femtoseconds to 10s of picoseconds interact with the environment via a wide variety of novel mechanisms, and the relevant coupling physics varies with pulse duration and wavelength. These systems offer a relatively unexplored landscape for directed energy applications, assuming energy transport from source to target is well understood. 

POA-24-RV-001: Utilizing novel data, fusion, and analytic techniques to enable Space Domain Awareness Superiority 

Closed: Review in process

Description
Space System’s Command (SSC) defines Space Domain Awareness (SDA) is as the ability to rapidly predict, detect, track, identify, warn, characterize, and attribute, threats to U.S., commercial, allied, and partner space systems. Numerous countries have tested direct ascent anti-satellite (DA ASAT) weapons against satellites in orbit, posing a military threat and creating safety hazards from debris fields. Due to the introduction of DA ASAT weapons against satellites in orbit, SDA is a key mission area in the defense of our nation’s commercial, allied, and military assets. For the USSF to protect satellites in geosynchronous orbit from an ASAT attack, threats must be detected, and decisions made quickly. Due to advancements in commercial SDA, a plethora of non-classified SDA data exist for university researchers to exploit and develop novel analytical techniques for Space Domain Awareness Superiority.

POA-24-RY-001: Near Infrared Sensing (NIRS)

Closed: Review in process

Objectives
General research topic to investigate smallest physical size electronics and optics required for continual detection of NIR signatures over a specified range while contending with additional in-band background energy. 

Approach
This research should investigate electronic (Focal Plane Array (FPA), discrete detector, etc.) and Optics (varying size, material, power schemes, etc.) to determine the smallest physical footprint that can be achieved to detect and track signatures provided by the GOV over a specified range within a specified FOV that will be provided after proposed contract award.

POA-24-RI-001: Enabling Data Efficiency for 3D Model Performance

Closed: Review in process

Description

Point Cloud Data (PCD) is a set of points 𝑃𝑖 with spatial coordinates 𝑃𝑖 = (𝑥𝑖, 𝑦𝑖, 𝑧𝑖) that represent an object. 3D PCD generated by LiDAR is accurate and precise, but high-density 3D PCD uses a lot of storage space and is unintuitive to process. Lowering the collection requirements may prove valuable in future model development.


This project topic is soliciting proposals that optimize the collection and/or generation of 3D point clouds.

POA-24-RI-002: Recommender System for Target Custody Prioritization 

Closed: Review in process

Background
In constantly evolving and varying environments, there is inherent risk in deploying objecting identification and tracking platforms. Additionally, rapid re-tasking of sensors requires awareness of dynamically-changing object environments in order to prioritize objects. Future Air Force operations require object tracking solutions in a time-sensitive context with varying priority and value of the objects. 


Previous and ongoing AFRL research topics have addressed the needs of object tracking at scale as well as optimization of resources. However, the existing work operates on the assumption that these practices will not be explicitly applied in highly varying environmental conditions in which the priority of a given object impacts may also be changing dynamically. This means rapid re-tasking is necessary to account for changes in the environment to mitigate performance shortfalls in the algorithms.

POA-24-RI-003: Verifiable Authentication for Data Analytics in Untrusted Clouds

Closed: Review in process

Background
Rank-aware queries such as top k allow users to specify ranking functions to score a database. An input to the functions produces each record a score, which can then be used for purposes such as ranking. While this feature makes them building blocks for data analysis and many other applications, rank-aware queries are known to be computation intensive to process, most having a complexity that is non-linear with respect to the number of records. As data continues to outgrow computer power, cloud services present a promising solution, especially to those who are resource constrained. With a small cost or even free, data owners can simply upload a database to a cloud and let it process queries on their behalf. Nevertheless, many people find it difficult to trust clouds in query processing. As a third-party, a cloud may intentionally provide wrong data such as an incomplete query result to save costs; as a data center, clouds are also ideal targets for hackers. This problem calls for research of the techniques to extend users with the capability to verify whether rank-aware query results they receive from a cloud are indeed correct.

POA-24-RI-004: Secure Video Over HF

Closed: Review in process

Background
Video transmissions consume vast resources in any domain. It is  especially true of RF. Most video feeds are high bandwidth VHF/UHF or satellite feeds.  HF data transfer focuses on text and small data packets. Novel techniques are  necessary to provide unidirectional video over such limited channels. 

Objectives
Secure video over HF requires efficient usage of very limited  resources. The main goal is to transmit 1 video frame per second unidirectionally without audio over an HF link. This link will be over a 2 KHz bandwidth channel. After  this a layer of security will be added, maintaining the same frame rate.  

POA-24-RI-005: Coalition AI Agent Collaboration Study

Closed: Review in process

Background
The DoD recognizes the need to advance DoD command & control (C2) to include but not limited to enhanced planning, wargaming, and course of action development to maintain/achieve decision dominance on the battlefield. One area of interest is utilizing the breakthrough in gameplaying artificial intelligence (AI) seen in the academic world through technologies including Reinforcement Learning, Computation Game Theory, Evolutionary Algorithms, Search, Optimization. Due to the nature of the DoD and our many partners and allies, a single monolith AI agent will almost certainly not be sufficient, instead AI agents will likely need to communicate and collaborate, while protecting their respective information. As the DoD looks to develop artificial intelligence (AI) approaches for complex adversarial environments, it is essential to develop capabilities to allow for federated or direct collaboration between AI agents that might be separated by command, domains, and nationality, which need to communicate and collaborate, while protecting their respective information.

POA-24-RI-006: Intent-Based Networking for Distributed C2

Closed: Review in process

Background
Future contested fights with a near peer adversary will demand the need to command and coordinate an increasingly dynamic set of time sensitive missions that are highly dependent on adversary, environment, and mission conditions. The evolving peer threat has created an operational reality that the Air Force must contend with; connected austere C2 nodes can be threatened and that communications avenues can be contested and degraded. This reality is driving the creation and adoption of Concepts of Operations (CONOPS) that are both agile and survivable by employing various adaptive basing strategies. These CONOPS have emphasized this inherent tradeoff between efficiency and resiliency in order to meet execution timelines, balance resource utilization, and account for the capacity of distributed operations.

POA-24-RI-007: Heterogeneous telecommunications quantum network

Closed: Review in process

Background
AFRL/RITQ aims to deliver heterogeneous quantum networking solutions to explore novel communication and networking protocols. To that end, we are constructing a classical and quantum network testbed at our open-campus facility, the Innovare Advancement Center. Once laboratory demonstrations are completed, successful technologies will be optimized for low-SWaP operation, with the goal of testing component technology outside of a well-controlled laboratory. A major near-term goal for the testbed is establishing entanglement distribution between heterogeneous quantum network nodes, including quantum integrated photonic nodes, superconducting qubit processors, and trapped ion atomic memory nodes. Network links will consist of standard telecommunications optical fibers and free-space optical links.

POA-24-RI-008: Analyzing Code Quality, Correctness, and Security of Large Language Model-Generated Code

Closed: Review in process

Background
Assurance of the resultant product of software development practices is paramount to the safety and security of modern digital technologies and must take precedence over all other concerns, such as development speed, as decisions are made on adoption of new practices. The latest paradigm shift in software development ushers in the innovation and automation afforded by Large Language Models (LLMs), showcased by Generative Pre-trained Transformer 3 (GPT-3), which has shown remarkable capacity to generate code autonomously, significantly reducing the manual effort required for various programming tasks. Although, the potential benefits of LLM-generated code are vast – most notably in efficiency and rapid prototyping – as LLMs become increasingly integrated into the software development lifecycle and hence the supply chain, complex and multifaceted challenges arise as the code generated from these language models carry profound questions on quality, correctness, and security. For example, zero trust attacks have emerged as the significant threat to supply chain integrity, the use of “Never Trust, Always Verify” approaches are called for to prevent threat actors seeking to exploit vulnerabilities in software components, now including those generated by LLMs. Research is required to comprehensively explore these critical concerns surrounding LLM-generated code, viewing them through the lens of software assurance.

POA-24-RQ-001: Multidisciplinary nonlinear time-spectral methods for air vehicle design 

Closed: Review in process

Objectives
To develop nonlinear time-spectral methods (e.g. harmonic balance) to include multidisciplinary coupling with sensitivities for air vehicle design optimization.

Description
Transient phenomena are foundational to air vehicle physics. It may be an intrinsic property of a system, such as a propeller wake impinging on a wing, or they may be an emergent phenomenon that manifests itself only in a multidisciplinary setting, such as flutter. In either case, it is important to account for such processes as early as possible in design to either mitigate adverse impacts to a vehicle or to formally leverage in design to the advantage of the overall vehicle. However, simulating transient processes is much more computationally expensive than steady-state processes and introduces additional challenges for calculating adjoint sensitivities for design; requiring management of both the forward and adjoint problems in time and their time-series data. Time-spectral methods (e.g. harmonic balance, nonlinear harmonics) seek to solve for a set of identified frequencies directly. As such, there is no need to march in time, which ameliorates challenges for data management, time-step resolution, length of time integration, and appropriate time-averaging for quantities of interest. The transient process must however be able to be represented by a discrete set of frequencies (e.g. flutter, turbomachinery flows, propeller airframe interaction). Of particular interest is the extension of nonlinear time-spectral methods for coupled-physics systems (e.g. fluid-structure) along with the calculation of their sensitivities for gradient-based design optimization. 

POA-24-RQ-002: Goal-oriented adaptive methods for air vehicle design 

Closed: Review in process

Objectives
To develop goal-oriented adaptive analysis methods for transient, coupled-physics phenomena and assess their application within processes for air vehicle design.

Description
The ability to quickly evaluate an air vehicle design realization during optimization is critical for the effectiveness of the design process overall. High-fidelity computational, coupled-physics analyses are an important tool for evaluating design objectives, but their relatively high cost can limit or prevent their early and widespread application in air vehicle design optimization. Goal-oriented adaptive methods seek to significantly reduce analysis time by reducing problem complexity while preserving predictive quality for outputs-of-interest (e.g. goals for design optimization). Goal-oriented adaptive methods that are of interest under this topic include mesh adaptation, h-p-r adaptation, space-time adaptive methods as well as supporting technologies such as error estimation. Of particular interest is the extension of such methodologies to include transient (time-marching and time-spectral) phenomena with multidisciplinary couplings. 

POA-24-RQ-003: Extending the Tractability of Large-Eddy Simulation for High Reynolds-Number Flows 

Closed: Review in process

Objectives
The proposed research will explore wall-modeled large-eddy simulation (WMLES) techniques for enabling tractable and reliable, first-principles-based simulation of high-Reynolds-number flows. The wall models will be implemented into an existing AFRL high-fidelity flow solver, FDL3DI, and their performance will be assessed for a range of test cases. Along with the models, several implicit time-integration schemes - beyond those currently available within FDL3DI - will also be implemented and investigated to further extend the stability and efficiency of the solver for relevant configurations and flow conditions.

POA-24-RQ-004: Analysis and Development of Inlet Systems for Gas Turbine Engines 

Closed: Review in process

Objectives
Future aerospace systems will span a wide range of speed and size classes. Each aircraft concept places a unique set of requirements on the integration of a high-performance inlet to feed air to the gas turbine engine. In our research group we seek to combine low-cost experimental testing with modern computational analysis to gain a better understanding of the flow field and enable advanced inlet concepts through a design, analyze, build, and test methodology. 

POA-24-RQ-005: Innovative Active Flow Control for Aircraft Aerodynamics Enhancement

Closed: Review in process

Objectives
Research and develop Active Flow Control (AFC) technology applications for the enhancement of Air Force aircraft aerodynamics. Potential applications for AFC solutions include: control of air refueling booms, wing separation control, fighter-type aircraft stability and control improvements, and the inclusion of AFC technologies at the conceptual design stage. Explore technology modeling at the right level of fidelity for the flow control device and aircraft aerodynamics to identify and develop relevant solutions. 

POA-24-RQ-006: Aerothermoelastic Analysis Methodologies for Aircraft Design 

Closed: Review in process

Objectives
To develop, apply, and assess computational methods for analyzing the physical interactions that result from coupling aerodynamics, structural dynamics, and heat transfer.  The methods should be sufficiently accurate to reliably capture nonlinear responses, while being sufficiently efficient to employ in a multidisciplinary design optimization procedure.  The methods should be applied to benchmark problems (e.g., wing structure or engine aft deck structure) of reasonable topological complexity and with realistic constraints (e.g., flutter). 

POA-24-RQ-007: Multi-Fidelity Analysis Methods for Multidisciplinary Design Optimization

Closed: Review in process


Objectives
Develop new computational procedures to predict the behavior of a coupled system in a collaborative manner using aerodynamic simulation methods at different fidelity levels.  


Description
A significant challenge in applying Multidisciplinary Design Optimization (MDO) to future aircraft systems is the high cost of performing accurate aerodynamic analysis, particularly in settings where there are important couplings with others disciplines (e.g., structure and control).  Strategies to lower analysis cost include various forms of physics-based reduced order modeling and more general forms of surrogate modeling, statistical analysis, etc. Another way to lower the cost of analysis is to introduce different modeling fidelities within a single computation; i.e., treat sub-domains at different analysis fidelity levels. A well-known example of this strategy is coupling an inviscid analysis method with a boundary layer analysis method to predict viscous flowfields.  What is desired in this topic is a more rigorous and general multi-fidelity approach across a wide range of fidelities (potential flow to Navier-Stokes).  Potential directions  include: goal-oriented adaptation (adjusting sub-domain fidelity requirements to minimize analysis cost while meeting accuracy targets for certain goals, such as drag); account of multidisciplinary couplings (e.g., new requirements levied by structural deformations, both static and dynamic), and more complex physics (e.g,. boundary layer transition, acoustics (near/far field), and thermal effects).

POA-24-RQ-008: High Frequency Load Measurements of Slender Bodies in Unsteady Flow Fields

Closed: Review in process


Objectives
Develop high frequency non-intrusive pressure measuring methods for measuring unsteady loads on slender body stores in an unsteady cavity environment.  


Description
To estimate a bound on store separation “trajectory spread” from weapons bays due to unsteady aero loads in the vicinity of the cavity, dynamic store balance measurements are required to accurately measure unsteady store loads in wind tunnels. A small cavity model scale increases forcing frequency applied to the store, while store/balance inertia limits traditional balance load response, which could result in inaccurate balance load measurements. A successful solution to this problem lies in a combination of intelligent design of the load measuring scheme, combined with an analytical approach to separate aero-dynamic loading from other non-aero loading picked up by the unsteady load sensor. High frequency non-intrusive pressure measuring methods (i.e. PSP) could be one way to estimate real-time unsteady aero loading on the store, but is typically limited by lack of a 360 deg view of the model. This could be supplemented by unsteady CFD to “fill in” missing views of the store for a complete 360 deg view, for integrated pressure loading over the whole surface. For the other aspect of the problem, unsteady CFD combined with a finite element model can be used to identify the unsteady test rig vibrations which will corrupt the load cells with non-aero forces. The ultimate result would be the development of a reduced order model, specific to a store / sting / cavity test setup, which could be used to back out aero-loads from total measured loads.  The Aerospace Systems Directorate program “Rapid Assessment of Weapons Separation (RAWS) is currently attempting to measure store load as they traverse thru a weapons bay flow field using traditional store balance techniques. Proposed work under this topic could consist of any or all relevant components of this problem – physics based computational, experimental, or low order modeling. Control techniques could be applied after the initial proof of concept, to enhance accuracy of measurement / minimize extraneous vibration. Ultimate application of the techniques is intended for larger wind tunnels (including the 2 ft by 2ft) TGF facility, but small university test rigs and sample problems are encouraged for initial proof of concept.

POA-24-RQ-009: Weapons Bay Shear Layer Physics Quantified with High Fidelity Computational Fluid Dynamics 

Closed: Review in process


Objectives
Develop high fidelity Computational Fluid Dynamics (CFD) model of a weapons bay cavity shear layer to aid in the understanding of the underlying physics. 

Description
Shear layer dynamics can play a significant role in determining the cause of store separation anomalies from weapons bays. Understanding the physics that governs the shear layer development over a weapons bay cavity is not thoroughly understood. Current experimental methods cannot provide the spatial and temporal accuracy to quantify the governing shear layer physics. High fidelity CFD can obtain the spatial and temporal accuracy to develop a better cavity shear layer model. The Aerospace Vehicles Division is attempting to develop a model that can predict the spatial and temporal flow dynamics inside a cavity and an accurate shear layer model is a critical piece of this effort.

POA-24-RQ-010: Detonation Propulsion Engine Research 

Closed: Review in process


Description
Detonation based propulsion systems are potentially revolutionary technology which utilizes the pressure rise of the detonative combustion process to produce momentum while adding heat with less entropy than conventional deflagrative combustion. Recent work at the Detonation Engine Research Facility at Wright-Patterson AFB has advanced the state-of-the art through computational and experimental studies utilizing several in-house developed codes and research engines, including pulsed detonation engines, rotary (continuous) detonation engines, and other pressure gain combustors. Areas of work include: detonation initiation, fuel injection, valving, controls, materials, heat transfer/thermal management, nozzles, ejectors, hybrid turbine engines, acoustics, power extraction, emissions, and diagnostics. A wide variety of research opportunities are available utilizing the unique detonation research facility, high speed instrumentation (up to 5 MHz per channel), high speed imaging (1,000,000 fps), laser diagnostics, and research engines.

POA-24-RQ-011: Advanced Multi-Scale Combustion Systems 

Closed: Review in process


Objectives
Perform research and development related to advanced multi-scale combustion systems useful for enabling future Air Force capabilities.


Description
Developing advanced concepts for future multi-scale (i.e., small- and medium-scale) combustion systems is important for many propulsion and power applications with significant impact and broad relevance to next-generation Air Force systems. Advanced multi-scale combustion systems provide the potential for enhancing the range, speed, and affordability of gas turbine engines.  The primary objective involves developing advanced multi-scale combustion systems concepts for small- and medium-scale gas turbine engines. 

POA-24-RQ-012: Rotating Detonation Engine Foundational Research 

Closed: Review in process


Objectives
Develop advanced tools and provide foundational knowledge useful for guiding the design and development of rotating detonation engines (RDEs).


Description
Improving foundational understanding of rotating detonation engines is important for many propulsion and power applications with significant impact and broad relevance to next-generation Air Force systems.  Rotating detonation engines provide the potential for enhancing the range, speed, and affordability of ramjet, rocket, and gas turbine engines.  The primary objective involves developing advanced experimental or computational tools and providing foundational knowledge useful for guiding the design and development RDEs.

POA-23-RI-001: Knowledge Graphs for Reasoning

Closed: This POA was modified and reposted but is currently closed.


Description:
Knowledge Graphs capture information about entities and the relationships between those entities, represented as nodes and edges within a graph. Entities can be comprised of objects, events, situations, or concepts. Knowledge Graphs are typically constructed from various data sources with diverse types of data, creating a shared schema and context for formerly disparate pieces of data. As such, Knowledge Graphs provide a rich source of information, enabling capabilities like question and answering systems, information retrieval, and intelligent reasoning.  Knowledge Graphs have been used to great success through companies like Google for internet search engine results, Amazon for product recommendations, and financial institutions for fraud prevention. Of interest to the Air Force are Knowledge Graphs that enable situational awareness, pattern of life analysis, and threat detection. 

POA-23-RI-004: Explainable AI for Reinforcement Learning

Closed

Description
As the Air Force begins to operate in contested environments against near-peer and/or peer adversaries, the demands on operational planners and warfighters will quickly increase, thereby requiring decision support assistance. In recent years, major developments in the field of Artificial Intelligence (AI) for video game playing agents has suggested that some of these approaches could be considered as candidates to provide that form of decision support. As the DoD looks to investigate and develop machine learning (ML) approaches for complex adversarial environments, understanding the underlying reasoning of these algorithms becomes essential especially when these tactics can have a direct impact on the warfighter.
   

POA-23-RH-001: Neural signals of uncertainty comprehension 

Closed: This POA was reopened but is currently closed.

Background
Decision making in the real-world is a complex task that involves, in part, the integration of information from multiple sources. Information sources can vary on their reliability in a variety of ways, however the uncertainty associated with any source must be communicated effectively to facilitate optimal decisions. This project seeks to identify the neural correlates that are associated with the comprehension of uncertainty information during decision making and evaluate how either endogenous (e.g., neuromodulation) or exogenous (e.g., uncertainty visualization techniques) manipulations impact uncertainty information comprehension. 


Objective
Collaborate with AFRL Cognitive Neuroscience researchers to design, implement, and analyze research identifying neural markers associated with uncertainty comprehension and develop intervention to facilitate optimal decision when incomprehension is expected/detected. 

POA-23-RY-007: Global Navigation Satellite System (GNSS) Library

Closed


Background/Statement of Objectives/Needs
General research topic to investigate automated, high gain RF collection and processing of L-band (950 MHz - 1650 MHz) GNSS signals to create an indexed data library for long term GNSS trust evaluations.

Approach
A satellite pass and collection schedule to be developed and maintained to ensure acquisition and revisit of all GNSS satellites visible from WPAFB over various azimuth and elevation angles several times a week. Once properly implemented, it is envisioned that the system will only require minimal, periodic (weekly) user input to validate/verify measurements.

POA-23-RI-002: Foundations of Self-Supervised Learning

Closed

Background
In Machine Learning (ML), Self-Supervised Learning (SSL) represents a growing set of learning strategies to pre-train ML models using unlabeled data. These models have shown tremendous promise for many downstream tasks, such as being fine-tuned for a classification task, utilized in low-shot learning, or even realistic image generation. The Air Force is interested in SSL to enable the creation of ML models in a label efficient manner and to aid in knowledge discovery.

Unfortunately, most SSL techniques have been developed for text and Electrical-Optical (EO) imagery (represented as single data modalities), or the pairing of EO imagery and text.  Undoubtedly, many Air Force applications are driven from EO imagery and text, but many critical applications also deal with additional data modalities. These include but are not limited to, Synthetic Aperture Radar (SAR), Ground Moving Target Indications (GMTI) from radar, the Electro Magnetic Spectrum (EMS), seismic sensors, and more. Although existing SSL techniques can be adapted for some of these data modalities, unique properties of these new data types are left unexploited by existing techniques. For example, SAR can be substituted for EO imagery in a contrastive learning SSL task, but existing image augmentation techniques ignore physical properties of SAR collection.

This research project is soliciting proposals to develop the foundational approaches to advance the understanding of SSL for Air Force specific challenges. At the link above are three high priority challenges on which proposals can focus, but we will consider sound proposals addressing challenges not listed here. Preference will be given to proposals grounded in operational Air Force data or reasonable facsimiles, and we encourage proposals to be well scoped to enable significant progress to be made on the specific research project. Please note that proposals should account for the data and compute resources needed to successfully complete the proposed work.

POA-23-RI-003: Interactive Learning for Mission Planning

Closed

Description
Many Air Force planning problems require the ability to quickly decide on an ideal action to take given some new information. Posing planning tasks as optimization problems allows for automation to provide the requisite speed. Challenges remain, however, in designing the optimization problems in the first place; it is usually difficult to translate human understanding of objectives into a mathematical language. This project will focus on making it easier to design and solve optimization problems that underlie various Air Force planning tasks, using interactive machine learning, new tools for designing automated planners, and a data collection framework that will allow for human responses. More specifically, the goals of this project are as follows:

POA-23-RI-001: Knowledge Graphs for Reasoning

Closed

Background
Knowledge Graphs capture information about entities and the relationships between those entities, represented as nodes and edges within a graph. Entities can be comprised of objects, events, situations, or concepts. Knowledge Graphs are typically constructed from various data sources with diverse types of data, creating a shared schema and context for formerly disparate pieces of data. As such, Knowledge Graphs provide a rich source of information, enabling capabilities like question and answering systems, information retrieval, and intelligent reasoning. Knowledge Graphs have been used to great success through companies like Google for internet search engine results, Amazon for product recommendations, and financial institutions for fraud prevention. Of interest to the Air Force are Knowledge Graphs that enable situational awareness, pattern of life analysis, and threat detection.  

This research project is soliciting proposals to develop the foundational approaches to advance the understanding of Knowledge Graph creation and exploitation for Air Force specific challenges. At the link above are three high priority challenges on which proposals can focus, but we will consider sound proposals addressing challenges not listed here.  Preference will be given to proposals grounded in operational Air Force data or reasonable facsimiles, and we encourage proposals to be well scoped to enable significant progress to be made on the specific research project.  

POA-23-RH-003: Airmen Stress Scenario Simulation using Cultured Cells: Mitochondrial Health & Organ-Level Effects

Closed

Background
Airmen constantly endure an evolving spectrum of operational stress scenarios that are vital to the continued success and achievement of the Air Force mission directives. These stress scenarios can include extreme physical exertion, high temperatures, excessive G-Force, pressure changes, low oxygen environments, or exposure to chemical or particle contaminants. Further, various stress factors induce changes in physiological or psychological attributes that affect performance through metabolic mediators that cause structural and functional recalibrations of mitochondria. Mitochondria are always in constant flux by changing their morphology and energy production in response to the energy (ATP) needs of the cell. Mitochondrial dysfunction is increasingly recognized as a contributor to premature fatigue and age-related muscle loss and functional impairment. The dynamic nature of the mitochondria allows for rapid detection of physical or cognitive impairment by characterizing the structure and the function of the mitochondria.

POA-23-RY-001: Machine Learned RF Waveforms 

Closed

Description
RF spectral analysis continues to be a challenging environment under which the sampling and sorting of various RF waveforms occurs. Commercially-based spectrum analyzers have productized continuous time-frequency waterfall displays enabling enhanced visualization, but still rely on a user interface to implement spot recording and detailed comprehension of the captured waveforms. With the expansion of spectrum allocation via 5G/6G telecommunications and the combinatorial possibilities of software defined RF waveforms with multiple features (i.e., Bi- Phase/Polyphase, pulse compression, modulation-on-pulse, pulse repetition interval, etc.), the demand for more computationally intensive signal processing methods needs to occur. The application of machine learning techniques like those associated with pattern and image recognition that could decipher and distinguish complex waveform characteristics under overlapping spectral and spatial domain conditions.

POA-23-RY-002: Rapid Geolocation of RF Emitters with Machine-learning

Closed

Description
Passive emitter location from a single airborne sensor platform is typically accomplished through an algorithm that tracks the convergence of strobes (i.e. rays originating from the location of the sensor) as direction of arrival measurements are accumulated. The underlying direction of arrival measurements are made in the presence of noise, which may be Gaussian (e.g. thermal noise) or impulsive (e.g. due to large inertial measurement errors). The application of artificial intelligence/machine learning is sought to accelerate the convergence of the geolocation solution by learning through the training process to ignore outlier measurements and to converge to the true location more quickly than conventional algorithms.

POA-23-RY-003: Data Mining of Analyst Processes for Task Automation 

Closed

Description
Data analysts utilize software tools that allow them to visualize/filter datasets in different ways in order to distill meaningful information from them. By logging numerous user interactions with a software visualization tool, new data can be generated that describes the process an analyst uses to arrive at some conclusion. The objective of this project is to utilize data mining and machine learning techniques to “learn” the steps and decisions an analyst makes when exploring datasets, and automate their process. The work is expected to take two years, and can be completed remotely.

POA-23-RY-004: Radar Association and Disambiguation using Graph Neural Network Link Prediction

Closed

Description
Passive emitter location from a single airborne sensor platform is typically accomplished through an algorithm that tracks the convergence of strobes (i.e. rays originating from the location of the sensor) as direction of arrival measurements are accumulated. The underlying direction of arrival measurements are made in the presence of noise, which may be Gaussian (e.g. thermal noise) or impulsive (e.g. due to large inertial measurement errors). The application of artificial intelligence/machine learning is sought to accelerate the convergence of the geolocation solution by learning through the training process to ignore outlier measurements and to converge to the true location more quickly than conventional algorithms.

POA-23-RY-005: Low Latency Wideband RF Signal Generation and Analysis Architecture for Closed-Loop Hardware-in-the-Loop (HITL) Simulators

Closed

Description
Current HITL architectures rely on a mix of analog and digital RF components. Every component, which the signal must pass through, adds data latency, frequency-based delays and additional noise. For an open-loop HITL simulator, the added delays and noise have minimal impact to the assessment of the system under test (SUT). This is not the case with closed-loop HITL simulators because it requires a return path to analyze the feedback of the SUT and provide the appropriate response. As the number of signals increase, the processing time using various resources increases thus reducing the appropriate response time. The request for this topic is conduct an analysis of alternatives of various architectures to minimize latency, delays and noise between processing and RF components in a closed-loop HITL simulator.

POA-23-RY-006: Polyglot Solver for Two Parsers

Closed

Background
A polyglot is a set of contiguous data that can be processed error-free by at least two parsers, where the parsers arrive at semantically different conclusions on the same data (e.g., the PDF parser and embedded filesystem created for and described in a series of articles in PoC||GTFO, Vol. 1).


Objective
The student will explore and develop mechanisms for solving polyglot problems.

POA-23-RY-008: Global Navigation Satellite System (GNSS) Trust

Closed

Background/Statement of Objectives/Needs
General research topic to investigate techniques and algorithms for establishing trust in GNSS signals by authenticating the source of transmission and/or computing integrity.


Approach
This research should build upon existing authentication and integrity techniques and be implementable in a software defined receiver (SDR) architecture. Use of external sensors should be minimized and the selected approaches should primarily use a combination of signal processing, statistical models, and machine learning to achieve GNSS trust.

POA-23-RY-009: Space Based Signals of Opportunity (SoOP)

Closed

Background/Statement of Objectives/Needs
General research topic to investigate positioning from space-based Signals of Opportunity (SoOP) that may include Low Earth Orbit (LEO), Medium Earth Orbit (MEO), and Geosynchronous Earth Orbit (GEO) commercial and military satellites. Both ranging and Doppler techniques should be considered.


Approach
This research should leverage a software designed receiver (SDR) architecture. Use of terrestrial based reference nodes and reliance on node-to-node communication should be minimized.

POA-23-RY-010: Terrestrial Based Signals of Opportunity (SoOP)

Closed

Background/Statement of Objectives/Needs
General research topic to investigate positioning from stationary land-based and sea-based Signals of Opportunity (SoOP) which transmit at L-band frequencies (950 MHz - 1650 MHz).


Approach
This research should leverage existing situational awareness systems to feed a software designed receiver (SDR) architecture. Use of terrestrial based reference nodes and reliance on node-to-node communication should be minimized.

POA-23-RY-011: AI/ML Geolocation of Ground-Based Laser Threats 

Closed

Background
U.S. aircraft are required to penetrate and operate in hostile environments. One of the major threats they face today are directed energy weapon systems. An ongoing concern is the efficacy of existing Countermeasure (CM) systems and techniques as these weapon systems and threat capabilities advance. Directed energy threat detection requires schemes containing multiple discriminants. The relative fidelity of coherent wavelength, direction-of- arrival, geolocation, fluence and pulse processing discriminants is dictated by the threat spaces of interest and drives the complexity of warning sensor architectures to produce the desired data products for the mission. This effort will be developing a reusable solution applicable to airborne geolocation of ground based laser threats to be implemented in multiple types of laser detection systems.

POA-23-RY-012:  Radar Behavior via Natural Language Processing Techniques

Closed

Description
As radar systems transition from hardware-limited to software-defined, single radar systems have become capable of intelligently selecting and interleaving multiple functions to detect and track numerous targets simultaneously. The goal of an electronic support (ES) system is to observe these behaviors and perform analysis to determine the intent of the radar system (e.g. is this a passive civilian radar, or a military system capable of guiding a missile) and predicting future behaviors. One proposed approach to this problem is to apply techniques from natural language processing (NLP). Just as NLP seeks to extract semantic meaning from sequences of letters which form words and sentences, the ES system seeks to extract radar intent from sequences of pulses which form bursts and dwells.

POA-23-RQ-001: Meta-materials for Structural Applications

Closed

Description
Micro-architected metamaterials, materials engineered at the sub-micron scale using tessellated or hierarchically arranged nano-lattices, exist in a previously vacant region of material performance space that enable non-linear, elastic structural response(s) in metals and ceramics. Though their potential is obvious and exciting, a tremendous amount of multi-disciplinary effort remains to realize their full capability at application relevant length scales. One such application, by way of example, is compliant structures such as gross airfoil shape changes (i.e. – span-wise twisting, leading edge manipulation, etc.) or continuous control surfaces (i.e. – flaps, ailerons, stabilators, etc.). Design concepts for these applications require massive amounts of strain (>10%) along specific directions while remaining stiff enough to support transverse aerodynamic loads at near-Mach relevant airspeeds. Micro-architected materials could provide one solution to this unique design space by enabling macro-scale material design that begins at the nano-scale. Due to this expansive tailor-ability, a validated modeling framework is necessary to efficiently explore the potential design space. Because macro-scale behavior is dependent upon millions of detailed, nano-scale lattices, reduced order modeling is necessary to ensure nano-scale effects are captured at the systems level. Near term, two specific challenges are a priority: 1) creating and validating a multi-scale modeling framework to understand the macroscopic (i.e. – meter-scale) material response of nanometer-scale design variables; 2) use of that validated framework to explore the sensitivity of macro-scale material performance to nano-scale fabrication resolution and defects. 

POA-23-RQ-002: Multidisciplinary nonlinear time-spectral methods for air vehicle design 

Closed

Description
Transient phenomena are foundational to air vehicle physics. It may be an intrinsic property of a system, such as a propeller wake impinging on a wing, or they may be an emergent phenomenon that manifests itself only in a multidisciplinary setting, such as flutter. In either case, it is important to account for such processes as early as possible in design to either mitigate adverse impacts to a vehicle or to formally leverage in design to the advantage of the overall vehicle. However, simulating transient processes is much more computationally expensive than steady-state processes and introduces additional challenges for calculating adjoint sensitivities for design; requiring management of both the forward and adjoint problems in time and their time-series data. Time-spectral methods (e.g. harmonic balance, nonlinear harmonics) seek to solve for a set of identified frequencies directly. As such, there is no need to march in time, which ameliorates challenges for data management, time-step resolution, length of time integration, and appropriate time-averaging for quantities of interest. The transient process must however be able to be represented by a discrete set of frequencies (e.g. flutter, turbomachinery flows, propeller airframe interaction). Of particular interest is the extension of nonlinear time-spectral methods for coupled-physics systems (e.g. fluid-structure) along with the calculation of their sensitivities for gradient-based design optimization. 

POA-23-RQ-003: Goal-oriented adaptive methods for air vehicle design 

Closed

Description
The ability to quickly evaluate an air vehicle design realization during optimization is critical for the effectiveness of the design process overall. High-fidelity computational, coupled-physics analyses are an important tool for evaluating design objectives, but their relatively high cost can limit or prevent their early and widespread application in air vehicle design optimization. Goal-oriented adaptive methods seek to significantly reduce analysis time by reducing problem complexity while preserving predictive quality for outputs-of-interest (e.g. goals for design optimization). Goal-oriented adaptive methods that are of interest under this topic include mesh adaptation, h-p-r adaptation, space-time adaptive methods as well as supporting technologies such as error estimation. Of particular interest, is the extension of such methodologies to include transient (time-marching and time-spectral) phenomena with multidisciplinary couplings. 

POA-23-RQ-004: Extending the Tractability of Large-Eddy Simulation for High Reynolds-Number Flows 

Closed

Description
The limiting factor in the application of large-eddy simulation (LES) techniques for Air-Force-relevant configurations and conditions is the extreme resolution required for resolving the near-wall regions of the flow with increasing Reynolds number.  Not only does this requirement dramatically increase the simulation’s computational size, but it also forces a proportionately reduced time-step for maintaining stability of the advancing solution, which leads to longer simulation time with Reynolds number for a set physical time. To overcome the severe scaling requirements, wall-modeled LES (WMLES) methodologies have been developed, in which the near-wall region is left intentionally under-resolved by the LES, but is solved through an auxiliary wall-stress model instead. Typical wall models are of an algebraic or ordinary differential equation form over the innermost boundary layer that couples to the LES by extracting velocities on its outer edge and feeds back shear-stress values to be used as boundary conditions on the LES at the wall. The model is oftentimes only formulated in the wall-normal direction and is much more computationally tractable to solve than the full three-dimensional Navier-Stokes equations in the same region. It is also not strongly tied to the time-step of the simulation, and thus breaks the severe spatial and time-step scaling relationship at the wall with increasing Reynolds number. This feature makes WMLES a promising methodology to enable tractable first-principle-based simulation for increasingly higher Reynolds-number, relevant conditions.

POA-23-RQ-005: Analysis and Development of Inlet Systems for Gas Turbine Engines 

Closed

Description
This research area covers the design and analysis of inlet systems for subsonic through supersonic applications. Aerodynamics of separated flow in diffusing internal ducts and shockwave boundary layer interactions are featured prominently in unsteady and three-dimensional configurations. Practical approaches to the use of computational fluid dynamics, which enable the minimum fidelity required for successful prediction of total pressure recovery and flow distortion, are needed to evaluate new inlet systems and explore aerodynamic flow control. This includes the use of hybrid RANS/LES techniques to compute unsteady characteristics of the distorted flow field at the fan/compressor interface.

POA-23-RQ-006: Analytic Sensitivities and Machine Learning applied to Uncertainty Quantification for Multidisciplinary Systems Analysis & Design Optimization

Closed

Description
Historically, Uncertainty Quantification (UQ) is performed late in the design cycle, when mitigation of deficiencies is costly or may result in a penalty to performance or capability.  These late defects and faults may be critical due to unanticipated interdisciplinary couplings or due to the uncertain nature (both aleatoric and epistemic) of anticipated interdisciplinary quantities of interest.  Types of uncertainty may include, but are not limited to:  parameter uncertainties, such as model or design parameters, geometric or material variables, and parameters associated with environment and process control; model uncertainties, such as from physics-based models from simple to complex, empirical models based on experiments, couplings/interfaces between disciplines, and model boundary conditions; data uncertainties, including noise, measurement errors, and missing data; requirements or usage uncertainty, including uncertainty in constraints; and, uncertainties arising from simulation, including discretization errors, round-off errors, and algorithmic errors.  Research opportunities exist in the general area of UQ, and in particular sensitivity analysis and machine learning applied to non-deterministic approaches. 

POA-23-RQ-007: Advanced Structural Concepts for lighter and low cost Aircrafts

Closed

Description
Modern aircraft has one of the most efficient aerodynamic shapes ever using advanced tools such as CFD. Aircraft structures and materials technologies have also developed over a hundred years since the first flight, but five or six decades old technologies are still widely used for substructure construction, which is composed of orthogonal structures such as spars, ribs, longerons, or bulkheads. These technologies could not meet endurance requirements due to heavy weight. For the last couple of decades, unitized structure and composite material based structure have demonstrated great potential to balance competing performance and cost requirements in aircraft design and in some instances, attain performance beyond traditional capabilities. AFRL is interested in exploring advanced structural technologies that enable light weight and low cost structure. Research opportunities exist in advanced structural concept design, low cost manufacturing, and design for assembly, maintenance, and repair. The topics of research include but are not limited to: (1) Structural optimization using topology optimization; (2) Innovative structural concept (e.g., morphing / reconfigurable mechanism or structure, bio-inspired structure, tensegrity structures, etc) ; (3) Multi-functional structures (e.g. antenna integrated structure, battery integrated structure, etc); (4) Design for manufacturing/assembly; (5) Advanced prototyping (e.g., continuous fiber composite, automated fiber placement, additive manufacturing, smart tools, etc) ; (6) Low cost manufacturing; (7) Enabling materials and actuators (e.g. flexible/corrugated materials, meta materials, smart materials, novel hybrid compact actuators); and more.

POA-23-RQ-008: Innovative Active Flow Control for Aircraft Aerodynamics Enhancement

Closed

Description
Based on progress in AFC in both device development and maturing applications, there is renewed interest in the understanding of potential AFC enhancements to Air Force aircraft aerodynamics. The most likely candidates for enhancement would be new aircraft developments, but opportunities to impact legacy systems may exist. Uses of the technology in aerodynamic design include the delay of flow separation for wing or control surfaces, flow control applied to an aerial refueling boom, and other applications for military aircraft. There is a need to better understand the effectiveness and costs of using AFC for a given application. Initial steps require establishing expected performance and requirements for AFC devices and then showing potential beneficial and detrimental impacts at a system level. 

POA-23-RQ-009: Enhanced Optical Flow Diagnostics / Minimization of Particle Seeding 

Closed

Description
One of the greatest challenges in implementing particle-based velocimetry techniques in air involves satisfactory flow seeding. Problems such as uniformity, density, particle size, satisfying Stokes’ criterion, matching characteristic time, tunnel access, and getting the seed to entrain into reverse flow regions are all considerable challenges.  All of these challenges are highly facility, condition, and model dependent.  Clearly a flow seeding technique that doesn’t require the use of Roscoe fluid, glycerine based seed, olive oil, DEHS, talc, coated glass balloons, phenolic resin and other materials (which clog ports, foul pressure sensitive paint, and obscure optical windows)  would be highly desirable.  Specific aerodynamic wind tunnel models of interest to test along with the optical velocity measuring technique will be specified based upon availability.  

POA-23-RQ-010: Aerothermoelastic Analysis Methodologies for Aircraft Design 

Closed

Description
The understanding of structural behavior under aerodynamic and thermal loading is relevant to many aerospace applications, including behaviors of structural components exposed to engine heating or heating generated by high aerodynamic speeds.   Thermal expansion and material softening degrade the structure, elicit static and dynamic responses (including loss of dynamic stability), and can complicate the prediction of these structural responses.  Structural heating is important to understand from several standpoints, including controllability (are control surfaces still effective?); engine integration (does vehicle shape-change impact propulsion?); stability (will heated wings flutter?), and sub-systems (do vital sub-systems get too hot over time?). 

POA-23-RQ-011: Multi-Fidelity Analysis Methods for Multidisciplinary Design Optimization


Closed

Description
A significant challenge in applying Multidisciplinary Design Optimization (MDO) to future aircraft systems is the high cost of performing accurate aerodynamic analysis, particularly in settings where there are important couplings with others disciplines (e.g., structure and control).  Strategies to lower analysis cost include various forms of physics-based reduced order modeling and more general forms of surrogate modeling, statistical analysis, etc. Another way to lower the cost of analysis is to introduce different modeling fidelities within a single computation; i.e., treat sub-domains at different analysis fidelity levels. A well-known example of this strategy is coupling an inviscid analysis method with a boundary layer analysis method to predict viscous flowfields.  What is desired in this topic is a more rigorous and general multi-fidelity approach across a wide range of fidelities (potential flow to Navier-Stokes).  Potential directions  include: goal-oriented adaptation (adjusting sub-domain fidelity requirements to minimize analysis cost while meeting accuracy targets for certain goals, such as drag); account of multidisciplinary couplings (e.g., new requirements levied by structural deformations, both static and dynamic), and more complex physics (e.g,. boundary layer transition, acoustics (near/far field), and thermal effects).

POA-23-RQ-012: High Frequency Load Measurements of Slender Bodies in Unsteady Flow Fields

Closed

Description
To estimate a bound on store separation “trajectory spread” from weapons bays due to unsteady aero loads in the vicinity of the cavity, dynamic store balance measurements are required to accurately measure unsteady store loads in wind tunnels. A small cavity model scale increases forcing frequency applied to the store, while store/balance inertia limits traditional balance load response, which could result in inaccurate balance load measurements. A successful solution to this problem lies in a combination of intelligent design of the load measuring scheme, combined with an analytical approach to separate aero-dynamic loading from other non-aero loading picked up by the unsteady load sensor. High frequency non-intrusive pressure measuring methods (i.e. PSP) could be one way to estimate real-time unsteady aero loading on the store, but is typically limited by lack of a 360 deg view of the model. This could be supplemented by unsteady CFD to “fill in” missing views of the store for a complete 360 deg view, for integrated pressure loading over the whole surface. For the other aspect of the problem, unsteady CFD combined with a finite element model can be used to identify the unsteady test rig vibrations which will corrupt the load cells with non-aero forces. The ultimate result would be the development of a reduced order model, specific to a store / sting / cavity test setup, which could be used to back out aero-loads from total measured loads.  The Aerospace Systems Directorate program “Rapid Assessment of Weapons Separation (RAWS) is currently attempting to measure store load as they traverse thru a weapons bay flow field using traditional store balance techniques. Proposed work under this topic could be consist of any or all relevant components of this problem – physics based computational, experimental, or low order modeling. Control techniques could be applied after the initial proof of concept, to enhance accuracy of measurement / minimize extraneous vibration. Ultimate application of the techniques is intended for larger wind tunnels (including the 2 ft by 2ft) TGF facility, but small university test rigs and sample problems are encouraged for initial proof of concept.

POA-23-RQ-013: Weapons Bay Shear Layer Physics Quantified with High Fidelity Computational Fluid Dynamics 


Closed

Description
Shear layer dynamics can play a significant role in determining the cause of store separation anomalies from weapons bays. Understanding the physics that governs the shear layer development over a weapons bay cavity is not thoroughly understood. Current experimental methods cannot provide the spatial and temporal accuracy to quantify the governing shear layer physics. High fidelity CFD can obtain the spatial and temporal accuracy to develop a better cavity shear layer model. The Aerospace Vehicles Division is attempting to develop a model that can predict the spatial and temporal flow dynamics inside a cavity and an accurate shear layer model is a critical piece of this effort. 

POA-23-RQ-014: CO2 Zeotropes Enabling Advanced Transcritical Thermal Cycles for High-Speed Systems 

Closed

Description
Supercritical CO2 has many beneficial, though nonlinear, thermodynamic properties near the critical temperature relative to either the liquid and/or vapor states. The pressure, and associated fluid density, near (and above) the critical point allows for significant SWaP savings. The inherent compactness of these systems enable a level of cooling and power generation capability not previously achievable on small sized platforms. This makes transcritical thermal cycles appealing due to the reduced flowrate requirements and size of system components.

POA-22-RX-003: Developing python-based hardware-software integration platform for closed-loop autonomous synthesis 

Closed 

Description
While recent advancements in artificial intelligence and machine learning have led to a number of technology breakthroughs, their application in basic scientific research has only started to emerge in the past few years. In the context of synthetic organic chemistry, traditional methods often utilize previous knowledge to perform iterative experiments to identify best experimental conditions for discovering new materials, where each iteration requires significant lead time. Developing computational tool sets that can directly interact and influence reaction outcomes as well as plan future experiments by utilizing AI/ML tools, accelerates the exploration and exploitation of experimental conditions to provide optimized yields of target molecules, polymers, or particles. The request for proposal is aimed at developing a python-based hardware-software integration platform to execute autonomous chemical reactions on an in-house AFRL platform via analytical decision making.

POA-21-RV-001: Leveraging Microgravity for Military/Commercial Applications & Products


Closed

Description
The recent explosive growth in the satellite and launch industries has led to a corresponding increase in microgravity processing interest and opportunities (e.g. Virgin Galactic, Space Station commercial opportunities, etc.). A host of commercial entities are now actively exploring uses of the microgravity environment in near-Earth orbit for commercial product applications, many of which have joint military applications. This topic will explore both potential dual use commercial/military microgravity products, as well as the fundamental science underpinning the microgravity processing environment.


POA-21-RV-002: Rocket Cargo Technology for Agile Global Logistics

Closed

Description
The Department of the Air Force is determining the viability and utility of using large commercial rockets for Department of Defense global logistics. We are interested in solutions to improve our ability to:

POA-21-RX-003: Predicting Infiltration Efficiency in Structures with Hierarchical Complexity

Closed

Description
Efficient manufacturing and consistent properties are the two greatest obstacles preventing ceramic and carbon composites from integration into supply chain and system design hierarchies. These challenges stem from the intricacies of infiltration and re-infiltration of a liquid precursor into a fibrous preform, and its subsequent pyrolysis to form the composite matrix. 

 

POA-22-RI-003: Developing Modular Neural Network for Intelligent Edge Computing

Closed

Description
As a powerful component of future computing systems, Deep Neural Networks (DNNs) are the next generation of Artificial Intelligence (AI) that intently emulates the neural structure and operation of the biological nervous system, representing the integration of neuroscience, computational architecture/circuitry, and algorithms. Overall however, DNNs still have limited architecture design perspectives in the following aspects: (1) The inefficient processing pipeline for a large-scale network structure; (2) The costly training operation with the increasing demand of data density; (3) The improper network behavior and diminished resultant accuracy with unknown objects.

POA-22-RI-001: Evaluating Assurance of AI-enabled Code


Closed

Description
BACKGROUND: This project will examine approaches across context spaces of: Problem-Definition Context (e.g. modeling, quantifying, and analyzing threats), Solution Definition Context (secure development, (semi-) formal, static, and dynamic analysis for vulnerabilities), and Requirements Context (trust assessment – was the correct system designed and built correctly) to reduce the introduction and/or exploitation of vulnerabilities in modern AI and AI-enabled systems.


OBJECTIVE: Collaborate with AFRL Assured Software researcher(s) to develop and validate new methodologies, techniques and tools to model and test against vulnerabilities in systems which include AI-enabled software.

POA-22-RY-001: Characterization of Defects in GeSiSn Thin Films and Developing Methods to Improve Material Quality for IR Applications


Closed


Description

For high enough concentrations of Sn incorporation, alloys of GeSiSn are the only group-IV materials which exhibit a direct band gap, making them attractive for potential integrated Si photonics applications in the IR.  However, the large lattice mismatch between GeSiSn alloys and Si substrates results in significant levels of strain and the formation of defects to reduce this strain.  To date, there have been few detailed investigations of defects in GeSiSn, their fundamental nature and their impact on devices fabricated from these alloys.

POA-22-RX-002: In Situ Micromechanical Study of Damage Evolution in CMCs

Closed

Description
Ceramic matrix composites (CMCs) are state of the art ceramic materials for high temperature applications. Continuous fibers embedded within a ceramic matrix allow for improved toughness capabilities compared to their monolithic counterparts. The mechanical behavior of these materials is strongly dependent on their microstructural features including fibers distribution, fiber coatings, and matrix porosity. In order to improve upon and understand the relationship between the CMC microstructure and overall performance, in situ microscale testing techniques can be utilized. In-situ microscale testing techniques, for these materials will focus on tensile testing and include in situ loading in the scanning electron microscope (SEM) (Figure 1a), in situ loading under an optical microscope (Figure 1b), and lastly in situ loading with micro computed tomography (CT) (Figure 1c). All of these capabilities at AFRL have different length scales with which to measure microstructural features. Each piece of equipment comes with their own capabilities: the SEM stage has the ability to do tensile loading and 4-pt loading at both room temperature and elevated temperature (1100°C). While the optical stage performs only tensile loading at room temperatures, it also has the ability to employ acoustic emission (AE) sensors to the sample and digital image correlation.

POA-22-RX-001: Predicting Damage Initiation and Propagation of Microstructural Experiments 

Closed

Description
The fundamental difficulty for predicting damage evolution in tough engineering materials involves extremely complicated nonlinear processes acting from the atomic scale through microscale and on up to the scale of the structure itself . Virtual testing, especially with respect to multiscale design, has been a topic of interest for those who design chemistries, understand processing, and ultimately performance. Fibrous composite materials can be broken down into multiple scales, where the scale of the constituents is of lower order than the scale of the resulting material and structure. The process of homogenization usually begins at the lower scales; however, some techniques have utilized macroscopic behavior to understand the material at a desired level of heterogeneity. Micromechanics has been helpful in determining the homogenized composite stiffness, or modeling damage, failure, or other nonlinear phenomena locally within the constituents. Understanding micromechanical phenomena is important because it has a significant influence on the cracking sequence and the statistics of crack initiation under any given load. However, even after extensive microscopic analysis being made based on macroscopic response, accurately predicting composite strength from constituent material properties (i.e, matrix, fiber, and matrix-fiber interfaces) remains a difficult task.

POA-21-RX-001: Generation and Evolution of Surface and Near-Surface Defects during the Processing of Ceramic and Carbon Composites

Closed

Description
Thermal methods occupy a major portion of the ceramic and carbon composites processing field. A significant challenge in obtaining reliable components lies in the unpredictability of the matrix microstructure formed during thermal decomposition of the polymeric precursor used in Polymer Infiltration and Pyrolysis (PIP) processing methods. Chemical models have been used to explain the compositional aspects of polymer-to-ceramic conversion, and a few emerging pyrolytic ones try to tackle bulk microstructure evolution. None, however, exist to address how this conversion is initiated at the beginning of the decomposition process, and more specifically what are the incipient morphological manifestations of the pyrolytic decomposition, and what is their distribution - either on the surface, or in the bulk of the composite. Since this is the process which initiates the creation of the pathways for environmental ingress, decomposition product evolution from the bulk, and subsequent precursor re-infiltration, it has a critical impact on the morphological formation throughout processing, and from there the final set of composite properties. This topic targets the understanding of the generation and evolution of near-surface porosity during pyrolysis of a ceramic composite in the green (i.e. cured) state. More specifically, it seeks elucidation of the surface and near-surface structural and compositional factors controlling the distribution, size and shape of the incipient surface defects initiated by the decomposition of the cured matrix precursor.

POA-21-RX-002: Developing environmentally friendly biosynthetic routes to synthesize aerospace-grade monomers and their precursors


Closed

Description
High-Temperature resin polymer matrix composites (PMCs) based components/elements are key in several exquisite AF platforms. However, the continued maturation and operational usefulness of these PMCs are often limited by the costs to procure and synthesize the monomers required by these resins. Either they are extremely expensive to synthesize, or their procurement is outsourced to foreign suppliers. The proposed request for proposals is aimed at reducing these barriers via exploring biological routes to synthesize aerospace monomers. 

POA-22-RI-002: Evaluating Software Assurance in an Agile Development Environment


Closed


Description

BACKGROUND: This project will examine approaches that could generate stronger empirical and formal analysis, be traceable to requirements, and be dynamically maintainable to keep pace with continuous agile development.


OBJECTIVE: Collaborate with AFRL Assured Software researcher(s) to develop and validate new methodologies, techniques and tools to model and test software security vulnerabilities in an agile development environment.

POA-21-RV-002: Rocket Cargo Technology for Agile Global Logistics

Closed

Description
The Department of the Air Force is determining the viability and utility of using large commercial rockets for Department of Defense global logistics. We are interested in solutions to improve our ability to:

POA-23-RH-002: In vitro proof-of-concept for engineered probiotics as vaccine alternatives to emerging viral pathogens

Closed: This POA was reopened but is currently closed.

Background
Respiratory infections are an active concern for the DoD, with respiratory infectious diseases accounting for up to 30% of related hospitalizations, and impacting up to 80,000 recruits and 600,000 active-duty service members each year; annually leading to 27,000 lost training days and 95,241 lost duty days [Agans, 2020]. The recent global SARS-CoV2 pandemic further highlighted, on a worldwide scale, how quickly pathogens can spread, become unmanageable, and dramatically impact operations and readiness. With the rise in monkeypox and polio cases, countermeasures and therapeutics which can be quickly developed, scaled in production, and disseminated to the widest market are increasingly needed....


Approach

Proposals for this effort should be related to engineering probiotics for vaccine alternatives. Proposals should address the following key goals:

Goal 1: Bacterial expression of immunogenic compound(s) on cell surface....  

Goal 2: Characterization of expression and resulting bacterial behavior...

Goal 3: Assess recognition of immunogenic target by immune cells....