Interpreting smell without relying on the cloud.
Robots need fast local decisions. Smell AI has to run on small hardware, on limited power, in the field — interpreting changing chemical patterns where they happen.
What we research
Mobile platforms cannot always wait for cloud analysis. Aeralyte targets on-device inference on the same low-power hardware that does the sampling — small models, compressed, running against a live sensor stream.
The pipeline
The same loop runs end to end: air is sampled, the sensor chamber responds, the response becomes a fingerprint, and an on-device model returns a label and a confidence — feeding a robot or IoT decision.
Where this stands
What is fixed is the design: one loop runs end to end on the same low-power board, with a hard false-alarm constraint on clean air built into the bar an on-device model must meet. And the input side is no longer hypothetical — the bench records controlled exposure streams for exactly this loop.
What is open is the part that matters: model behavior on those recorded streams — nothing is claimed until it clears the truth gates — and the latency, memory, and power budgets of running inference on the ESP32-S3 itself, measured on the same low-power board that does the sampling.
Further reading
- →Machine smell — the array and the fingerprint the model reads.
- →Drift & reality — keeping the model honest as the world changes.
- →Truth gates — the evidence bar every model result must pass.