OUSD (R&E) CRITICAL TECHNOLOGY AREA(S): Integrated Network System-of-Systems
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 develop and train cutting-edge machine learning models for edge deployment via TAK using the Model Integration Software Toolkit (MISTK) format.
DESCRIPTION: Training can be accomplished server-side, but inference must be done on device. TAK-ML, a client and server-side framework for ML development, and NodeDrop, a technology to reduce the size of neural networks without affecting efficacy, are provided to performers. Sample models/algorithms developed in and integrated with TAK-ML are provided (e.g., biometrics, edible plants). Example use cases may include, but are not limited to geolocation, command and control, search and rescue, surveillance, communications, IOT, cloud or intelligence (including open-source intelligence). Use of digital engineering tools to at a minimum define the APIs and where applicable build reference implementations is preferred. Leveraging TAK-ML and StreamlinedML to integrate into the TAK ecosystem is strongly preferred.
PHASE I: This topic is intended for technology proven ready to move directly into a Phase II. Therefore, a Phase I award is not required. The offeror is required to provide detail and documentation in the Direct to Phase II proposal which demonstrates accomplishment of a “Phase I-like” effort, in this instance demonstrating familiarity and proficiency with applied machine learning, preferably at the tactical edge.
PHASE II: As an applied ML topic, Phase II objectives mirror standard machine learning lifecycle steps to include data collection, model architecting and design, implementation either standlone or via registration/integration with provided AFRL toolkits, training, testing, and evaluation at the tactical edge.
PHASE III DUAL USE APPLICATIONS: Successful Phase II technology development will be eligable for additional Phase III work, with specific transition paths depending on the domain and problem set selected by the proposer. AFRL will work with the Tactical Assault Kit (TAK) Product Center (TPC) and domain-relevant end-user communities to promote transition of machine learning models that reach sufficient TRL (5-7) and interface well with mobile end-user devices in use by operators in the field.
KEYWORDS: mobile machine learning; end-user devices; edge computing; machine learning/artificial intelligence; resource constraints