SBIR Topic Watch Search topics Pricing Browser plugin Sign in

DoW/USAF · DoW SBIR 2027 BAA

DAF27BZ02-DV017 — Low - Cost Neuromorphic Edge Intelligence for Persistent Autonomous Change Detection and Machine Situational Memory

Pre-Release | released 2026-11-04 | opens 2026-12-02 | closes 2027-01-06 | CMMC Level 2 (Self), SBIR

SBIR.gov page | DSIP topics app (search DAF27BZ02-DV017)

TPOC: Joseph Fernandez <[email protected]>; Isabel Zaru-Roque <[email protected]>

Keywords: Neuromorphic Sensors, Event-Based Vision, Change Detection, Edge AI, Persistent ISR, Machine Memory, Event Cameras, Autonomous Sensing, Low-SWaP Computing, Multi-UAS Collaboration

Technology areas: Air Platform, Electronics, Sensors

Objective

Develop, integrate, and demonstrate a low-SWaP-C neuromorphic IR sensor to support celestial navigation to enable position and navigation of sUAS in a Global Navigation Satellite Systems (GNSS) denied environment in day and nighttime for world-wide operations. The desired system should allow an autonomous aircraft to establish a persistent machine representation of an area, recognize operationally meaningful changes, and communicate primarily what changed rather than continuously transmitting full-motion imagery or raw sensor data.

Description

Traditional airborne ISR architectures frequently collect and transmit large quantities of data even when little or nothing within the observed environment has changed. This creates substantial burdens on: Communications bandwidth. Processing. Storage. Power. Analysts. Command-and-control systems. Neuromorphic and event-driven sensors offer a fundamentally different sensing architecture in which the system responds primarily to changes within the observed environment. The desired capability should combine event-driven sensing with onboard edge intelligence and temporal machine memory. Fundamental engineering considerations include: Event-driven/neuromorphic sensor characterization. Low-latency onboard processing. Low power consumption. Detection of operationally relevant changes. Rejection of irrelevant changes caused by lighting, weather, vegetation, shadows, sensor motion, or environmental clutter. Temporal comparison across repeated observations. Persistent representation of previous environmental state. Confidence scoring for detected changes. Autonomous prioritization of significant changes. Ability to cue additional onboard sensors. Generation of compact change products instead of continuous raw-data transmission. Multi-UAS sharing of relevant environmental changes. Open interfaces for integration with aircraft mission systems. Desired change classes may include movement, addition, removal, modification, displacement, construction, damage, altered activity patterns, or other Government-defined changes.

Phase I

As this is a Direct-to-Phase II (D2P2) topic, no Phase I awards will be made. To qualify, applicant(s) must provide Feasibility Documentation demonstrating prior completion of Phase I-equivalent efforts outside SBIR funding. Offerors should demonstrate feasibility through existing neuromorphic sensing, event-driven vision, edge AI, machine perception, autonomous change detection, or related capabilities. Relevant evidence includes: Prototype sensors. Hardware demonstrations. Airborne testing. Edge-processing demonstrations. Power/SWaP measurements. Change-detection performance. False-alarm characterization. Relevant commercial deployments. Preferred maturity is TRL 5–6, although compelling TRL 4 technologies may be considered when the underlying sensor and processing architecture has been demonstrated and can reach operational relevance within approximately three years.

Phase II

Task 1 – Sensor and Edge Architecture Integrate event-driven sensing, processing, storage, autonomy, and aircraft interfaces into a representative airborne package. Task 2 – Autonomous Change Detection Develop algorithms capable of distinguishing meaningful environmental change from irrelevant temporal variation. Task 3 – Machine Situational Memory Develop persistent representations enabling the system to compare current observations with prior observations across meaningful time intervals. Task 4 – Autonomous Cueing and Prioritization Allow detected changes to autonomously cue additional sensors or higher-fidelity processing and prioritize information for transmission. Task 5 – Multi-UAS Situational Memory Demonstrate the ability for multiple aircraft to contribute to and consume a shared representation of detected environmental change. Task 6 – Flight Demonstration Demonstrate the capability in an operationally relevant airborne environment. Performance should quantify: Probability of detecting relevant change. False-positive/false-change rate. Detection latency. Processing requirements. Power consumption. Bandwidth reduction. Data-storage reduction. Performance against environmental nuisance changes. Performance across repeated observations. Desired Phase II End State A TRL 6+ airborne prototype capable of autonomously identifying, prioritizing, storing, and communicating operationally relevant environmental changes while demonstrating measurable reductions in bandwidth, processing, storage, power, and/or analyst burden.

Phase III / dual use

Successful technology is intended to provide AFSOC, ACC, and operational organizations a rapidly fieldable sensing architecture capability for future electronic-warfare, communications, sensing, processing, and self-protection payloads. The transition owner is PAE ISR SOF / Asymmetrical Warfare Division – WISS. The architecture is intended to support future adaptation to additional platforms through further integration, development, and minimized cost burden. Follow-on funding is exploratory from AFSOC based on D2P2 results.

References

  1. Bovolo, F., & Bruzzone, L. (2007). A Theoretical Framework for Unsupervised Change Detection Based on Change Vector Analysis in the Polar Domain. IEEE Transactions on Geoscience and Remote Sensing, 45(1), 218-236.
  2. Chen, Y. H., Krishna, T., Emer, J. S., & Sze, V. (2017). Eyeriss: An energy-efficient reconfigurable accelerator for deep convolutional neural networks. IEEE Journal of Solid-State Circuits, 52(1), 127-138.
  3. Gallego, G., Delbruck, T., Orchard, G., Bartolozzi, C., Taba, B., Censi, A., Leutenegger, S., Davison, A. J., Conradt, J., Daniilidis, K., & Scaramuzza, D. (2020). Event-based Vision: A Survey. IEEE Transactions on Pattern Analysis and Machine Intelligence, 44(4), 154-180.
  4. Lichtsteiner, P., Posch, C., & Delbruck, T. (2008). A 128× 128 120 dB 15 µs Latency Asynchronous Temporal Contrast Vision Sensor. IEEE Journal of Solid-State Circuits, 43(2), 566-576.
  5. Mead, C. (1990). Neuromorphic electronic systems. Proceedings of the IEEE, 78(10), 1629-1636.
  6. Radke, R. J., Andra, S., Al-Kofahi, O., & Roysam, B. (2005). Image change detection algorithms: a systematic survey. IEEE Transactions on Image Processing, 14(3), 294-307.
  7. Verhelst, M., & Moons, B. (2017). Energy-Efficient and Edge-AI processors for the IoT. 2017 IEEE International Electron Devices Meeting (IEDM).

Copied from the official source and may be out of date. Confirm dates and requirements in the official solicitation.