SBIR / STTR #DON27TZ02-NV002

Navy STTR research on efficient, explainable machine learning for edge hardware

The Department of the Navy is seeking a research partnership to develop field-friendly, explainable machine-learning systems that can run accurately on resource-limited devices. - Government buyer: Department of the Navy, administered by the Office of Naval Research (ONR) and the Navy SBIR/STTR Program Management Office. - Topic: DON27TZ02-NV002, “Field-Friendly Super Intelligence and Machine Learning Systems,” within a Navy STTR solicitation. - Requested work: Develop an explainable, reliable, resource-efficient machine-learning model with image-classification performance comparable to deep neural networks (DNNs), while reducing computation and memory needs. It should run on CPUs and edge hardware without relying on cloud services. - Development phases: Phase I studies feasibility and prioritizes convolutional neural network (CNN) image classification; Phase II develops and tests a prototype against defined performance metrics; Phase III targets an industry-ready, scalable model, extending work toward transformers, multimodal data, and potential uses such as autonomous robots. - Notable requirements: Phase I generally has a six-month base and a six-month option; the Phase I funding caps are $200,000 for the base and $115,000 for the option. The topic seeks a partnership between a small business and a research institution. No product quantities or part numbers are specified. - OEMs and vendors: No commercial OEMs, vendors, or named research partners are identified. - Broader notice context: The accompanying Navy STTR material covers 15 research topics, including the NV-002 topic and work in optics, autonomy, energy, marine monitoring, manufacturing, antennas, and photonics.

Description

This is a Department of the Navy STTR topic under the DoW STTR 2027 BAA, administered by the Office of Naval Research. The source identifies the topic as pre-release and provides Navy proposal submission instructions, including a Phase I technical volume limit of 10 pages and submission through the DoW SBIR/STTR Innovation Portal. The topic question deadline is January 23, 2027, and proposals are due February 6, 2027.

View original listing