Embedded Machine Olfaction for Multi-Modal Sensing

This STTR Phase I effort proposes a biologically inspired electrochemical and photonic olfaction system designed to detect and classify aerosolized chemical and biological threats in real time. Developed by RealNose.ai in collaboration with MIT Media Lab, the platform integrates stabilized mammalian olfactory receptors into functionalized sensor arrays, supported by a machine learning engine capable of embedded classification under low power and size constraints. The project will validate sensor sensitivity (target LOD <10 ?g/mł), analyte selectivity under interferants, classifier accuracy on resource-constrained hardware, and performance during mobility and airflow perturbations. Building on prior NIH-supported work in medical diagnostics and DARPA-aligned sensor development, this effort adapts RealNose’s proven biosensing framework for tactical, mobile defense deployment. Phase I will culminate in a validated benchtop prototype with fully documented sensor performance, embedded classifier results, and motion-tolerance benchmarks. These results will directly support transition to Phase II system integration with mobile robotic or handheld detection platforms, meeting key objectives outlined in CBD25D-TP001.