META_Attachment_7_-_Statement_of_Objectives_(SOO).pdf

Attachment 7

STATEMENT OF OBJECTIVES (SOO)

Multi-Sensor Exploitation for Tactical Autonomy (META)

I. Background

The Multi-Domain Sensing Autonomy Division (RYA), Sensors Directorate (RY), Air Force Research Laboratory (AFRL), conducts research, exploratory and advanced development focused on battlespace understanding necessary to close the loop around multiple, multi-domain mission effects chains (MECS). RYA is responsible for providing multi-domain sensing autonomy technologies for intelligence, surveillance and reconnaissance (ISR); Strike; electronic warfare (EW) and Cyber; and command and control (C2) systems. This effort will support research to enable high confidence detection, ID, tracking and pattern of life (including behaviors and intents) of critical mobile targets (CMTs) in contested environments; provide flexible decision making algorithms, and deliver autonomous exploitation algorithms to augment analysts and enable closed-loop operation. This effort builds upon previous signal processing/exploitation efforts that focused on the extraction of information from physics-based single sensor data.

This effort will include knowledge generation through information fusion of multiple information sources and provide solutions for contested environments when data is extremely limited. This includes developing new technologies to address challenges related to multi-domain sensing for strategic, operational and tactical autonomy. This research effort will provide techniques for behavioral/physical knowledge generation from multiple sensors, domains (Air, Space, Cyber), and information types to include modeling, simulation, algorithm development, assessment, demonstrations and experiments across multiple distributed, homogeneous and heterogeneous sensors. The knowledge generated will support and inform the Observe-Orient- Decide-Act (OODA) loop for strategic (system and intelligence preparation for missions), operational (mission planning) and tactical (mission execution) elements of a single integrated multi-domain battlespace. The timeliness of the generated knowledge determines its ability to contribute to a strategic (long-term) plan, e.g., ISR, or to a tactical (short-term) decision, e.g., Strike. This research will focus on technologies including fusion for understanding (using spatial diversity, frequency diversity, transmit and receive adaptivity and information source diversity), target phenomenology characterization, denied and difficult target prosecution, algorithm performance assessment, experimentation, sensing decision-making strategies, knowledge representation, sensing data and knowledge management, distributed processing and capabilities to inform/enable joint inference and control. High performance computing (HPC) will be leveraged for data generation, modeling, simulation, algorithm development and performance assessment, demonstrations and experiments as appropriate.

II. Scope

The scope of this effort is to develop novel sensing autonomy technologies. Research objectives involve knowledge generation through information fusion, using multiple sensors, INTs and domains to create “mission essential knowledge”, defined here to mean detections, tracks, IDs, patterns, behaviors and intents. This research effort shall also investigate the application of machine/deep learning techniques to accomplish knowledge generation. The effort is estimated to be 15% basic, 70% applied and 15% advanced research. The projected TRL’s for the various research efforts range from TRL 1 to TRL 4. In addition, while not included in the BAA Technical Objectives in Paragraph III below, there is a potential that this effort, which includes any contracts issued resulting from this BAA, may also include technical objectives for incidental Operations and Maintenance (O&M) work and funding and incidental Procurement work and funding.

III. Technical Objectives

Technical objectives include, but are not limited to:

a.) Hard and Soft Information Fusion: Utilize combinations of both physics-based

“hard” information (e.g., sensors, INTs) and human-based “soft” information (e.g., social media, human-generated reports) to create mission essential knowledge (i.e., detections, tracks, IDs, patterns, behaviors, intents).

b.) Multi-Level Fusion: Fuse information sources at multiple levels (e.g., decision, feature, signal) to create mission essential knowledge (i.e., detections, tracks, IDs, patterns, behaviors, intents). Information sources can be collocated or distributed;

passive or active; and cooperative or non-cooperative.

c.) Machine Learning: Apply machine learning to multiple sensors, INTs and domains to create mission essential knowledge (i.e., detections, tracks, IDs, patterns, behaviors, intents) for military applications. Investigate use of transfer learning, deep learning and other modern methods for knowledge generation in contested environments.

d.) Autonomy: Develop flexible approaches/techniques toward autonomous/automated reasoning, with respect to creating mission essential knowledge, e.g., approaches that enable peer-to-peer coordination, cognitive operations and task flexibility.

e.) Data for Algorithm Development, Testing and Evaluation: Data for this effort includes the collection of measured data using airborne platforms and/or data predicted through simulation codes, which provide synthetic data representative of measured data. Measured, predicted and/or representative data will enable fusion and ML algorithm development and algorithm performance characterization. This includes data truthing and data wrangling. Data truthing is the characterization of the content of image or signal data in terms of real-world objects and events, and their position(s) within the data stream. Data wrangling is the process of normalizing data formats in preparation for data analysis; this includes characterization of meta-data and the operating conditions under which data was collected or obtained. Data wrangling also includes the formation of intermediate exploitation products which are then used for additional technology development, e.g., the generation of synthetic aperture radar (SAR) imagery.

f.) Data Alignment and Uncertainty: Develop techniques for space-time alignment, and information uncertainty modeling, of multiple information sources to enable determination of mission essential knowledge (i.e., detections, tracks, IDs, patterns, behaviors, intent). Information sources can be collocated or distributed; passive or active; and cooperative or non-cooperative.

g.) Limited Training Data: Develop approaches to generate mission essential knowledge (i.e., detections, tracks, IDs, patterns, behaviors, intent) in environments given assumption of limited (measured) information from any single source, i.e., contested environments. This involves development of algorithms that are robust to limited data availability. It also includes supplementing measured training data with synthetic data, and the investigation of how to close the gap (statistically) between these two types of data (measured and synthetic). This can include ML-based approaches such as transfer learning.

h.) Explain-ability: Investigate approaches to improve explain-ability of machine learning algorithm outputs/decisions to build/improve human trust.

i.) Satellite Applications: Explore sensing autonomy and sensing exploitation concepts (ML-based and fusion) for multi-level space to surface ISR architectures including National Technical Means (NTM), MILSPACE, commercial and allied country constellations.

j.) Commercial Leverage: Develop approaches for rapid application/leverage of and assessment of commercially-available machine learning techniques with respect to military problems of interest.

k.) Performance Models: Characterize algorithm performance through modeling, simulation and/or experimentation over various operating conditions (OCs) with an emphasis on contested environments.

l.) Novel Exploitation Products: Develop novel exploitation products, e.g., 3-D or super-resolution imagery, which enable and/or improve knowledge generation (i.e., detections, tracks, IDs, patterns, behaviors, intents).

m.) Demonstrations and Testing: Display, exhibit and validate performance and timing of novel approaches for behavioral/physical knowledge capture/generation through ground-based or airborne flight demonstrations/testing.

IV. Management

Management includes maintaining clear government visibility into program schedule, performance, and risk while allowing the contractor maximum flexibility and innovation to manage the effort. This includes program schedule, cost, performance, risks, associate contractor agreements (as needed), subcontracts and data to meet contract objectives that satisfies all technical requirements.

a.) Meeting: Host a kick-off meeting within 30 days of contract award. Program management reviews and technical interchange meeting will be required. Agenda for all meetings will be developed jointly.

b.) Incidental Material, Supplies and Testing Assets: Procurement of materials if they are not available through government channels and are required for the performance of this program. The materials shall consist of Materials may include consumable and non-consumable test equipment, supplies, replacement and new parts, components and subassemblies related to any of the sensing and processing elements, manufacturer's literature, and manuals which are not available through government channels and/or not available in sufficient time to meet the urgency of requirements. Flight/ground testing requirements may involve the need for and use of contractor-owned/operated and/or government-owned/operated testbed aircraft.

V. Research Deliverables

Research deliverables include technical reports, research results, computer software prototypes and related computer software documentation, hardware prototypes and related documentation and algorithm performance characterization over various operating conditions (OCs). Computer software format to be source code. Contract Data Requirements List (CDRL) Items: See BAA attachment.

VI. Flight Test (if applicable)

Comply with all the requirements of the AFRL test planning and approval process. This includes but not limited to Technical Review Board (TRB) and Safety Review Board (SRB) reviews, AFRL airworthiness reviews in addition to all Air Force policies, regulations, and instructions related to flight testing. Coordinate spectrum requirements for all META testing and conduct any required coordination with the test location (e.g. range requirements, airspace coordination, ground instrumentation, etc.)

VII. OPSEC Statement for AFRL/RY Contracts

General Operations Security (OPSEC) procedures, policies and awareness are required in an effort to reduce program vulnerability from successful adversary collection and exploitation of critical information. OPSEC will be applied throughout the life cycle of the contract. The Critical Information List (CIL) will be provided upon request by AFRL/RYOY Information Protection Office. While working on the government installation, OPSEC guidance will be provided by the AFRL/RYOY Information Protection Office.

VIII. Program Protection Plan

All DoD contractors (including subcontractors) shall supplement their current security practices by requiring any personnel involved in executing this contract where critical program information (CPI) has been identified to protect the CPI to the standards articulated in the Program Protection Plan and in accordance with DoDI 5200.39. Upon contract award, all identified DoD contractors (including subcontractors) shall acknowledge and meet the requirements stated by the Program Manager for the protection of CPI. The DoD contractor must immediately notify the U.S. Government upon the discovery of any nonconformance with CPI protection.