6 of 6 boards
Sensor values
Any slowly changing signal: current, temperature, pressure, vibration. Teach a board to notice when a reading drifts away from normal, to name the state a machine is in, or to estimate one value from the others.
Limits first
- Data arrives by CSV import only; there is no live capture for these signals yet.
- The downloaded code runs the model. Reading your sensor and filling the model's input is your code.
Which boards can do it
| Board | Support | Why |
|---|---|---|
| ESP32-S3-DevKitC-1 (N8R8) | Supported | |
| Arduino Nano 33 BLE Sense Rev2 | Supported | |
| Raspberry Pi Pico 2 W | Supported | |
| Nordic nRF52840 DK | Supported | |
| ST Nucleo-H753ZI | Supported | |
| ST Nucleo-F446RE | Supported |
How you collect it
Your phone cannot record these, so you bring them in as a CSV file from a data logger, a spreadsheet or your own board, and map its columns to readings, label and recording.
What you need
- To spot unusual readings: normal readings from at least 3 separate sessions.
- To name states: at least 5 recordings of each state, from at least 2 separate sessions, before the report will quote an accuracy.
- To predict a value: at least 3 sessions, with the value given for each recording.
Getting a good model
Anomaly detection and value prediction both work from window statistics: short summaries of each stretch of readings that are cheap to compute on the board.
Start a sensor values project
Create a free account, open a project and add your first examples.