Most countertop toasters are simple appliances that operate a set of heating coils on a timer. If you use different types of bread, thicknesses, or starting temperatures without changing the timer, you might end up with undercooked or worse, burnt toast. I tackle the elusive problem of making the perfect toast by applying some embedded machine learning trickery.
I modify (nay, hack) an inexpensive toaster to control the toasting process. By using a collection of gas sensors, I can collect abundant data on different odors released during toasting. If we can smell when toast is burned, why can’t a machine do the same?
I start by building a sensor collection cage that mounts a number of sensors over a toaster. From there, I burn a mountain of toast to collect enough gas data to train a machine learning model that can predict the time remaining before the toast becomes burned. This model is used to control the toasting process and stops the process some time before the burned stage is reached.
If you’d like to build your own, a full tutorial can be found here: https://www.digikey.com/en/maker/projects/how-to-build-an-ai-powered-toaster/2268be5548e74ceca6830bf35f0f0f9e
This project was inspired by Benjamin Cabe’s AI Nose project, which you can read about here: https://blog.benjamin-cabe.com/
Previous project (AI nose): https://www.youtube.com/watch?v=KyMC0LsLZms
Intro to TinyML Part 1: https://www.youtube.com/watch?v=BzzqYNYOcWc
Intro to TinyML Part 2: https://www.youtube.com/watch?v=dU01M61RW8s
