Edge AI with MicroPython and TensorFlow for Microcontrollers

In this presentation I will show different machine learning models that can be executed on a small microcontroller. For this purpose, I train the model in a Jupyter notebook and demonstrate how to integrate it on the microcontroller. As microcontroller I use a Raspberry Pi Pico with MicroPython and an integrated TensorFlow package.

The talk will include examples of implementing AI on a small microcontroller, including:

• Implementing a wake word system to wake up devices without constant data transfer to the internet
• The recognition of handwritten numbers by the microcontroller

I will also discuss the advantages and disadvantages of Edge AI and TinyML on a microcontroller.

‍