/dance-activated-microbit

Machine learning project for micro:bit that recognises TikTok dance moves using on-board accelerometer

Primary LanguageCMIT LicenseMIT

Dance activated micro:bit

A demo application that makes your micro:bit V2 model respond to TikTok dance moves, built with Edge Impulse. This demo uses Machine Learning to analyse the accelerometer data coming from the micro:bit on-board sensor, showing a smiley on the display when it recognises "Floss" TikTok moves.

Dance activated micro:bit

Project

Detailed instructions for the project are available on the OKdo Project Hub

  1. Clone this repository:

    git clone https://github.com/LetsOKdo/dance-activated-microbit

  2. Change to project directory:

    cd dance-activated-microbit

Build using Docker (tested on GNU/Linux Debian 11)

  1. Install Docker for your host Get Docker

  2. Build the container:

    docker build -t microbit_dance_detector .

  3. Build the project:

    macOS, Linux

    docker run --rm -v $PWD:/data microbit_dance_detector

    Windows

    docker run --rm -v "%cd%":/data microbit_dance_detector

Build natively (tested on GNU/Linux Debian 10 with gcc-arm-none-eabi-9-2020-q2-update)

  1. Install CMake, Python 2.7 and the GNU ARM Embedded Toolchain and Ninja. Make sure arm-none-eabi-gcc and the other components are in your PATH.

  2. Build the project:

    python build.py

Flash the image:

  1. Plug in your micro:bit to the host with a micro USB cable (the microbit should be visble as a mass storage device in your filemanager)

  2. Flash the binary to your micro:bit by dragging MICROBIT.hex onto the MICROBIT disk drive or copy the one from this repo.

Dance

  1. Detach your micro:bit and connect it to the battery pack

  2. The center LED on the matrix will light up showing that the micro:bit is ready to detect your dance moves

  3. Start dancing the "Floss dance" and if you do it right the matrix will light up with a smiley face

  4. The detection will reset automatically after a few seconds

How to recognise new dances

You can build new models using Edge Impulse.

  1. Sign up for an account and open your project.
  2. Install Edge Impulse tools from the command window
npm install -g edge-impulse-cli

  1. Set up one micro:bit as a dance move transmitter

  2. Set up another micro:bit as a dance move receiver

  3. Follow the Data Forwarder instructions

  4. Go back to Data acquisition and now collect your dance move data.

  5. Follow these steps to train your model.

    Note: use window length 999 instead of 1000!

Once you've trained a model go to Deployment, and select C++ Library. Then:

  1. Remove source/edge-impulse-sdk, source/model-parameters and source/tflite-model.
  2. Drag the content of the ZIP file into the source folder.
  3. If you've picked a different keyword, change this in source/AccelerometerInferenceTest.cpp.
  4. Rebuild your application.
  5. Your micro:bit now responds to your own TikTok moves 🚀.

There's an excellent video tutorial on using Edge Impulse here:

Voice-activated micro:bit