SBIR Topic: N242-092 — GigEVision-compliant Event-based Cameras · Sponsor: U.S. Navy · Phase I complete

Event-based sensors offer things conventional cameras physically cannot, but they speak their own language: a sparse, asynchronous event stream that standard machine-vision software and infrastructure are not built to ingest. This effort closed that interoperability gap.

Note: Some details have been altered, obscured, or sanitized to safeguard operational security and proprietary methods.

Our Approach

With Alwyn Smith, P.E., as Principal Investigator, our engineers led the embedded-hardware and systems integration that puts commercial event sensors on the standard machine-vision interfaces fielded systems already use: Ethernet, GigE Vision, GenICam, and 10GigE. The work covered event-sensor integration, data packaging, and standards-compliant transport across hardware, firmware, software, power, and mechanical design.

The Technology

The platform pairs commercial event-based image sensors with embedded edge processing on AMD K26-class system-on-modules. It delivers more than 120 dB of dynamic range and onboard AI/ML inference. Phase I culminated in an end-to-end demonstration of event data streaming over 10GigE to standard GigE Vision host software, with sensor latency tuned to roughly 120 µs, well inside the 2 ms per-payload requirement. The platform stood at TRL 5/6 at the close of Phase I.

Why It Matters

Standards-compliant transport is the difference between a laboratory sensor and a transition-ready one. Event-based detections can now drop into existing toolchains without re-engineering the downstream pipeline.

Project Images