Description:
OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Advanced Materials
The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the statement of work (SOW) tasks intended for accomplishment by the FN(s) in accordance with the Announcement. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws.
OBJECTIVE: The objective of this topic is to integrate a Mobile Device Manager (MDM) with a machine learning (ML) management infrastructure (StreamlinedML, WingmanAI) to allow the ability to define, compare, evaluate, train, deploy, and update ML Models for use in Mobile Machine Learning (MML) applications which support the TAK ecosystem in a secure, reliable environment for both cloud-based and disconnected environments.
DESCRIPTION: The Tactical Assault Kit (TAK) ecosystem currently has no end-to-end means to evaluate, compare, fine tune, and deploy MML to end user devices at scale. The environment should enable the comparison of models (e.g., Model Cards) with onboard inferencing. Provide the ability to scan ML models to check functionality, security, and reliability. Provide enterprise management of a “Marketplace” for ML models within the MDM’s “App Store”. Support both connected and disconnected environments. Provide the ability to review and version control, models and apps/plugins before analytics are pushed back to the developers on how the models are used. StreamlinedML, a government ML management and TAK-ML, a TAK-oriented ML development framework, are provided to performers. Use of digital engineering tools to at a minimum define the APIs and where applicable build reference implementations is preferred.
PHASE I: This topic aims at D2P2 awards with a "Phase I-type" minimum feasibility study that demonstrates experience developing, deploying, orchestrating, integrating and managing the likes of a Tactical Assault Kit (TAK) with a Mobile Device Manager in conjunction with a machine learning (ML) management infrastructure framework that supports the evaluation, comparison, fine tuning and deployment of Mobile Machine Learning to end user devices at scale.
PHASE II: The phase II objective of this topic seeks to integrate a Mobile Device Manager (MDM) with a machine learning (ML) management infrastructure (StreamlinedML, WingmanAI) to allow the ability to define, compare, evaluate, train, deploy, and update ML Models for use in Mobile Machine Learning (MML) applications which support the TAK ecosystem in a secure, reliable environment for both cloud-based and disconnected environments. StreamlinedML, a government ML management and TAK-ML, a TAK-oriented ML development framework will be provided to performers. Use of digital engineering tools to at a minimum to define the APIs and where applicable build reference implementations is preferred.
PHASE III DUAL USE APPLICATIONS: Successful Phase II technology effort reaching suitable TRL (6-7) will be candidates for additional Phase III development, including potential for transition to the Tactical Assault Kit (TAK) ecosystem in partnership with the TAK Product Center (TPC). In addition, Phase III efforts will focus on delivering the TAK mobile device manager technology with a machine learning (ML) management infrastructure to potentially a broader speactrum or series of diversed customers for operational use in a relevant commercial/civilian, or government/military working environment.
REFERENCES:
- https://tak.gov;https://civtak.org;
- https://dl.acm.org/doi/abs/10.1109/MILCOM52596.2021.9652909;
- https://github.com/raytheonbbn/tak-ml;https://github.com/mistkml/mistk
KEYWORDS: mobile machine learning; machine learning model verification; mobile device management; machine-learning marketplace; machine-learning model cards
