Technology · Neuromorphic computing
Computing only where something happens.
Since our founding we have worked with methods modelled on nervous systems: event-driven and frugal. That lets AI run where there is no mains power and no data connection.
Proven in projects
- Time-domain classifiers instead of an FFT, powered by harvested vibration (VCD4.3)
- Neuromorphic pattern matching on infrared thermography (MYW.AI 4 QM)
- Neuromorphic FPGA processing in the EOSC Smart-RNA project, code open on GitHub
Where it helps
- Sensors powered by vibration instead of a battery
- Assets without a data connection, read out over RFID
- Very small processors on a tight power budget
Next step
Little power, no connection?
We check whether on-sensor evaluation is possible for your asset.