6 of 6 boards
Motion
Gestures and movement, from an accelerometer or your phone's motion sensors. Teach a board to tell a wave from a shake, notice when a machine starts vibrating differently, or estimate how hard something was tapped.
Limits first
- Your board needs a motion sensor. The Nano 33 BLE has one onboard; every other board needs a sensor module wired to it.
- The downloaded code runs the model. Reading the sensor and filling the model's input is your code, and the example shows where it goes.
Which boards can do it
| Board | Support | Why |
|---|---|---|
| ESP32-S3-DevKitC-1 (N8R8) | Supported | Module |
| Arduino Nano 33 BLE Sense Rev2 | Supported | Onboard |
| Raspberry Pi Pico 2 W | Supported | Module |
| Nordic nRF52840 DK | Supported | Module |
| ST Nucleo-H753ZI | Supported | Module |
| ST Nucleo-F446RE | Supported | Module |
How you collect it
Hold your phone and press Record, or switch on continuous recording and the take is split for you wherever the phone keeps still. On the board, the same signal comes from an accelerometer module or an onboard motion sensor.
What you need
- To name movements: at least 5 recordings of each, from at least 2 separate sessions, before the report will quote an accuracy.
- To spot unusual movement: recordings of normal movement from at least 3 separate sessions. A few examples of faults help, but they are not required.
- To predict a number, such as how hard a tap was: at least 3 sessions, with the number typed against each recording.
Getting a good model
Record each movement in several short sessions rather than one long one, and vary how you hold the phone a little. The report holds whole sessions back for testing, so a model that only knows one session cannot hide it.
Start a motion project
Create a free account, open a project and add your first examples.
Other signals: SoundSensor values