Attachment 5 - Task Order 1 Statement of Objective.pdf

Statement of Objective (SOO)

PROJECT NAME:

• Distributed Data Automated Management System (DDAMS)

PROJECT PRIMARY POC:

• Contract POC:

• Technical POC:

PROJECT BACKGROUND:

The Distributed Data Automated Management System (DDAMS) is a solution to improve data processing capabilities by reducing the manpower required to organize data, prime data for use in advanced data processing techniques, and provide datasets for use in system evaluations, support model development, and advance Decision Driven Data (DDD), Figure 1. DDAMS uses Machine Learning (ML) and Deep Learning (DL) in a cloud environment to sort, tag, score, and process lower-level data to support additional post-processing, data analysis, reporting, and AI/ML training datasets. Through the implementation of DDAMS the cycle time for test decisions is significantly condensed, post test reporting time-lines are significantly reduced, and increased quality of the analyses/evaluations are realized due to improved availability of collected data. Further, the availability of historic test data to be efficiently queried and utilized for more comprehensive trend analyses can be achieved. This capability will be deployed on the Army Test and Evaluation Command (ATEC) – Yuma Proving Ground (YPG) cArmy environment (Microsoft Azure Cloud) to conduct data management and processing of a wide variety of test data types that include data from test items, range instrumentation [video/stills, Time Space and Position Information (TSPI), analog, and digital data], and environmental conditions. The scale and scope of this project is limited to Air Delivery Systems testing at YPG and is focused on the system architecture, database design, and enable further growth of this capability as a comprehensive solution tailored to ATEC’s enterprise-wide requirements.

Figure 1 - ASET Distributed Data AI Management System (DDAMS) Capability Data Flow

PROJECT DESCRIPTION:

DDAMS implmentation for the Air Delivery Systems capability at YPG will utilize modern technologies in distributed processing, ML, and DL to automate raw data management processes to support other processes, not developed under this project, such as data reduction, results analysis, reporting, and to conduct analysis of larger / longer time-span data and formatting data for implementation into AI training data sets.

The DDAMS, Figure 3, is the proposed solution to provide the following impacts for ASET and is scalable to a YPG, ATEC enterprise and broader DOD T&E solution.

• Reduce the workload of data management.

• Provide cloud distributed data access utilizing SQL/noSQL/hybrid databases

• Automate the sorting and classification of data

• Enable an environment for AI-ML-DL data processing

• Enable data collected to be scored for use in AI system training in the ATEC-YPG cloud environment.

Additionally, for testing at Air Delivery Systems the data collected is common across all the various test capabilities and DDAMS is intended to sort, score, and process data from its raw form that is uploaded to the cloud [cAmry] and have it available quickly, but not in real-time, to support technical analysis using AI, ML, and other custom data processing capabilities that are specific to

YPG.

Figure 3- ASET Distributed Data AI Management System (DDAMS) Flowchart

The concept of DDAMS for Air Delivery Systems is to address pinch points in workload and processes such that test results are available and accessible in a timely manner to support on-going testing and quick reaction results. The main pinch point for testers at the tactical edge is data management. After a test a tester will spend >50% of their time just getting data into the right location, assigning data to the correct event, truncating large data sets to the area of interest, and waiting for data processing that is dependent upon disparate data sources. This high percentage of time to just get the data in the correct configuration significantly limits the ability of the tester to complete the analysis and provide meaningful conclusions in the report. These challenges will be addressed by DDAMS leveraging AI-ML-DL to process raw data and assign it to the test events or areas of interest.

As an example, test video data could be uploaded into a general directory in the YPG Impact

Level 5 (IL5) environment on cArmy where the DDAMS cloud processing capability would reside.

The DDAMS would recognize that a file has been uploaded, utilize meta data to generally assign the video to a test program, then use ML computer vision capabilities to recognize what was in the video [aircraft, parachute, vehicle, payload,…], read the time of the video (on the video or embedded), and then assign the segments of video to the event(s)s that occurred for a test and make the data available in an noSQL/hybrid database.

Another portion of DDAMS is the implementation of the database to manage the storage, processing, searching, and access of test data. The current plan is to utilize a hybrid Structured Query Language (SQL) database to support the structured data generated while testing [such as tabular configuration data] and a Not Only SQL (noSQL) database to support the vast unstructured meta data that is read and/or assigned by DDAMS. This approach will be scalable to all the collected digital data [video, audio, sensor, network traffic, etc.] at the test range while also supporting human generated meta-data like test officer notes. The database developed for DDAMS will be designed to interface with the ATEC Common Reference Model (CRM) such that it seamlessly interfaces with the ATEC Data Mesh

The DDAMS improves timelines, responsiveness to failures, and provide a comprehensive view of a test that is not currently available. Other pinch points with similar impacts are the ability to conduct ML-DL analysis on very large, multi-modal, data sets to understand interactions, long-term [years] trend analysis of system performance, and the ability to support system development by providing labeled data for AI training. With this data, testers will have the ability to quickly provide results, have more time to understand and report on the implications of results, and provide improved and timely conclusions on system performance. The overall intent of DDAMS is to reduce the workload spent in the planning and analysis phases of a test to allow more time and man hours on the conduct and reporting on a test.

The database developed for DDAMS will be designed to interface with the ATEC Common

Reference Model (CRM) such that it seamlessly interfaces with the ATEC Data Mesh

GOVERNMENT PROVIDED:

• ATEC/YPG will coordinate with the contractor in order to provide example data files of typical data and applicable reference data.

• The government will have 60 days after delivery to review the software and provide comment / questions.

• ATEC/YPG will provide data on available architecture and provide access to SME’s

REFERENCE DOCUMENTS:

• None

KEY CHARAHTERISTICS:

• Deployed on the ATEC-YPG cArmy environment IAW ATEC Cloud Strategy (Microsoft

Azure)

• Not an infrastructure solution, deployable onto enterprise-wide cloud solution

• Is not tailored to a specific test item or capability, is focused on data collecting across the enterprise and is scalable beyond the Air Delivery Systems test capability

• Not intended for real-time analysis, but reduce data processing timelines

• Focused on priming data for analysis tools and development

• Limited scale to support Air Delivery Systems capability to improve ATEC Cloud Strategy and requirements

• Project will utilize existing capabilities to develop a comprehensive solution that is tailored to

ATEC TECHNICAL enterprise-wide requirements

• Process 100% list Air Delivery data type by sorting/scoring/tagging primary types with 95% accuracy

• Utilize ML capabilities to automate data pre-processing to support automated data processing and availability of data to users

• Implement on local YPG computer(s) and in U.S. Army Test and Evaluation Command (ATEC) Impact

Level (IL) 5 cArmy environment to enable broad access and reduced sustainment

• Accept raw & Engineering Unit (EU) converted input data as a source for flexibility

• Reduce data availability delay, from test event to distribution for analysis by test officer, by 50%(T) and

80%(O)

• Utilizes Modular Open Systems Approach (MOSA) to enable additional new data types and scalability

CRITIAL PARAMETERS

• Support input of YPG standard videos, System Under Test (SUT) configuration data, processed

GPS/IMU files, processed 1553 Bus, and processed analog / digital.

• Automatically sort and assign data input to individual test events.

• Reside on cArmy ATEC-YPG environment for data storage and remote processing.

• Open architecture that enables addition of new data types, data storage growth, and addition of new automated processing capabilities.

• Provide reporting and visualization of data available for each test event.

• Searchable by test event and system configuration

• Labeled data exported in standard format to support AI system training.

• Graphical User Interface that is intuitive and predictable

List of Data Input Types for Air Delivery Systems DDAMS

• High-Definition Ground to Air Video

• High-Definition On-Board Aircraft Video

• High-Definition On-Board Airdrop Payload Video

• Processes Differential GPS and IMU Text File

• Airdrop Payload Processed Analog Sensor Text File

• Aircraft Payload Processed Analog Sensor Text File

• Aircraft Processes 1553 Bus Text File

• Ground to Air Video Derived Time Space and Position Information (TSPI) Text File

• Still Photographs

• Test Configuration Tabular Data

• Data Collector Notes Tabular Data

• Environmental Conditions Tabular Data (Wind, Temp, Density,…)

DEVELOPMENT STATEGY:

The project is laid out in three phases. The first is to refine the system requirements, architecture, and design to ensure that scope is managed, deployable requirements are developed, and risk is reduced early in the project. The first test capability area to be incorporated into DDAMS would be Air Delivery.

This was chosen not because of its importance or higher need for a capability, but due to the lower number of system of systems interactions, the simpler data sources, and the unclassified nature of the work. This will enable DDAMS developers to manage a smaller and simpler case prior to developing a broader capability. The last phase of development will be to incorporate DDAMS into the Aviation Systems Branch and the Sensor Systems Branch, which have similar data types, interfaces.

TECHNICAL APPROCH: Developed by Contractor

GOVERNMENT REQUIRED MILESTONES OR MEETINGS:

Monthly technical interchange meetings.

• Monthly technical interchange meetings. On-site technical support of software integration

• Quarterly written progress reports

• Monthly technical interchange meetings

• Every 6 months or IOC/FOC a technical demonstration of capability

• Installation, training, and 2-year support / update service

REQUIRED PROJECT DELIVERABLES / CAPABILITIES:

• DDAMS Software (all source code, supporting documentation IAW industry standards, unlimited government rights)

• On-site technical support of software integration

• Quarterly written progress reports

• Monthly technical interchange meetings

• Project review meetings every 6 months or IOC/FOC a technical demonstration of capability

• Installation, training, and 2-year support / update service

• Process 100% Air Delivery sorting/scoring/tagging primary types with 95% accuracy

• Utilize ML capabilities to automate data pre-processing to support automated data processing and availability of data to users

• Implement on local YPG computer(s) and in U.S. Army Test and Evaluation Command (ATEC)

Impact Level (IL) 5 cArmy environment to enable broad access and reduced sustainment

• Accept raw & Engineering Unit (EU) converted input data as a source for flexibility

• Reduce data availability delay, from test event to distribution for analysis by test officer, by

50%(T) and 80%(O)

• Utilizes Modular Open Systems Approach (MOSA) to enable additional new data types and scalability

VALIDATION:

The government shall have 60 days after delivery to provide comment and ask questions on software and presentation content.

ASSUMPTIONS:

• New capability, not dependent on existing procure technology or capabilities.

• Will leverage existing capabilities to reduced development costs for an ATEC enterprise solution

PROJECT SCHDULE: Developed by Contractor

PERIOD OF PERFORMANCE: Developed by Contractor