ADN
TinyML closes the first edition of Arduino Unchained
Tiny Machine Learning, LoRaWan and other innovative technologies have shaped the new Arduino course. Plenty of news, in line with the latest market trends.
On Thursday 13 June we wrapped up the first edition of our brand-new training course Arduino Unchained, a programme of six intense evening sessions devoted to technological innovation.
It was the second new format in our training offering, following the two editions of Contemporary Python, and as usual it sold out and built a long waiting list — confirming the great interest in digital technologies.
As always, helped by the convenient evening time slot, the audience was extremely diverse: from young students to industry professionals, ages ranging from 15 to 70, all sharing curiosity towards new technologies and the desire to experiment.
Arduino is the platform that democratised microcontroller technology, and its classic boards — by now an icon of the maker movement for almost twenty years — continue to be the favourite of anyone approaching this world for the first time.
Yet our course aimed to explore the latest frontiers of possible applications. On Arduino, by now, the internet has it all: the supply of free, high-quality content is huge, and it’s hard to propose a professional course that tells something truly new.
Arduino is increasingly attentive to the professional market, and its product range has been enriched with new boards and new communication options.
That’s why we decided to introduce, alongside the foundational topics, a set of new, important themes for which there isn’t much documentation yet on the web.
This was the case for the LoRaWan protocol and for scalable, affordable IoT architectures. In the course of a single lesson our students built an end-to-end application from scratch, from reading a barometric sensor to sharing its data on a professional cloud console, using industrial-class products and platforms.
But above all, it was Thursday 13 June’s session that allowed us to explore one of the most promising research areas of recent years: Tiny Machine Learning on microcontrollers.
It’s a technology born in 2018 that started to develop between 2020 and 2021, and then was overshadowed by the explosion of Large Language Models, which became synonymous with Artificial Intelligence.
But the ability to use Machine Learning on microcontrollers is quietly revolutionising the market. A new generation of increasingly smart devices, that don’t need to be constantly connected to the network, is enabling the development of high-performance, scalable edge computing systems.
The synergies between the new Generative AIs and networks of devices running Deep Learning firmware on board will shape the technology landscape of the near future.
The topic is very recent and there isn’t much introductory content out there yet. For the occasion we redefined the TensorFlow Lite for Arduino library, creating a new release aligned with the latest version of Google’s general TensorFlow platform.
Students on our course had the opportunity to try it as a preview and experience first-hand how simple and immediate the use of this new technology can be when working with Arduino tools.
Machine Learning was presented as a development process alternative to traditional methods and just as easy to use. An additional, extremely powerful tool that lets us solve otherwise intractable problems.
Now that Arduino has been “unchained”, we’re preparing a flood of new content to explore and deepen these new topics.
We’ll meet again after the summer break for the start of the new training season, which will be rich in news.
The Arduino Unchained course is available from today, online, in the e-learning version on the new Accademia delle Arti Numeriche platform, organised in six content-rich Missions, with over 30 hours of high-resolution video tutorials.