▲ Click on the blue text above to follow us and not miss any valuable articles!
Since its establishment in 2005, the Arduino open-source platform has completely changed the electronics engineering field, with innovation represented by open-source sweeping the entire technology development landscape. Whether in the past, present, or future, the goal of technology has always been to make technology simpler and easier for people to use. This philosophy, along with Arduino’s open-source principles, has remained unchanged in its development plans.（Original link: https://www.eeworld.com.cn/aifDaLO）
We saw this in a recent interview with the company’s founder Massimo Banzi, and Arduino CEO Fabio Violante confirmed it as well.
Fabio Violante, CEO of Arduino
Violante told MIT Technology Review Italy: “We are investing in various areas to make technology accessible to more people. One of them is simplifying the software development process. Today, thanks to artificial intelligence, we can provide code development support not only to those with a certain level of knowledge but, most importantly, even to those without electronic design experience. To this end, we are adding AI capabilities to our development environments, including cloud and offline environments, as well as third-party products originating from Arduino.”
What else does the impact of artificial intelligence cover besides programming?
Our strength has always been microcontrollers. Today, thanks to artificial intelligence, smart sensors have become a particularly important area for Arduino.
A trend emerged a few years ago regarding the implementation of neural accelerators within microcontrollers. This means that various tasks can be performed while maintaining the typical characteristics of microcontrollers: low power consumption and the possibility of starting in microseconds (thus not having to wait for startup like computers or objects with microprocessors), such as voice assistants in home automation.
We are working to leverage the advantages of microcontrollers enhanced by neural accelerators to create smart sensor applications, such as low-power smart cameras. Currently, we have launched Nicla Vision on the market, thanks to our collaboration with OpenMV, which allows access to a vertical development environment in MicroPython to create AI-based machine vision applications. We are not talking about competing with products like Nvidia’s Jetson, but rather providing the possibility to perform simpler operations, such as activating insect traps in agriculture. This is an important application in the field, as it allows you to limit the amount and quantity of pesticides needed. All of this is achieved through a very small object (the size of a stamp) equipped with a camera and a very powerful dual-core microcontroller, Wi-Fi, Bluetooth, and a LiDAR module, along with an integrated digital microphone.
Nicla Vision allows you to listen to insect noises through a digital microphone, and when AI recognizes the patterns of harmful species, it takes a photo and measures the shape of the weight sample, thanks to a special scented bag that attracts insects. Once this data is collected, it is sent to the cloud. These applications were unimaginable a few years ago due to their size and power consumption, but now they can be realized with smart microcontrollers.
Therefore, besides the hardware aspect, software development has also received a lot of attention. How do you research innovations to integrate into your systems, especially regarding artificial intelligence?
The AI software modeling industry is very complex, and it is not our core business. This is why we rely on external partnerships, such as our partnership with Edge Impulse, the market leader in this field. We have used their system, which allows AI models to be easily integrated with microcontrollers and adapted to our systems. For example, for Nicla Voice, we implemented a specific neural network model in the hardware, making the device extremely efficient in terms of power consumption, and directly equipped it with a hardware accelerator on a dedicated Syntiant chip. But there are many applications: these neural networks can also be used for digital signal processing and many other tools, making it very easy to update models and capture the training set of neural networks (of course, this requires a lot of resources).
What are your expectations for the development of artificial intelligence? Looking ahead, how do you think AI will be implemented in future technological systems?
What we see today is the development focused on training these models. Centralized architectures dominate, such as ChatGPT, Anthropic, etc., with huge investments in data centers. Interestingly, the new generation of processors, especially Arm-based processors, can provide more powerful computing capabilities in low-power systems. It can be said that current smartphone processors have 40-50 TOPS of computing power. This allows very complex processes (such as those related to robotics) to be executed directly on the device. In the future, I envision robotic systems helping people and households, and law masters solving problems by interpreting people’s voices directly at home, without relying on external clouds. This is a trend that is already happening: efficient, fast, and low-power technology implemented directly in processors can manage very complex AI models. Therefore, processors and microcontrollers are transitioning from cloud to edge, and the dream of edge computing, especially edge AI, is approaching. The latter will be able to provide easy-to-use tools and expand the crowd of people who can unleash their creativity to create solutions to new problems, including speed, privacy, and data security. Building a system that fully aligns with our dream of technological democratization.
Making these technologies available to more users also means generating more data that can be further used for the development of other services and tools?
It is precisely for this reason that, like Arduino, we are investing in another area, namely facilitating the collection and processing of data (for example, data from sensors), and then training AI models. This process allows us to quickly train multiple models, verify which models can work, filter them, test them to create Arduino sketches (the code that can be loaded and run on the boards) and add them to our microcontrollers. We start with AI because it is the best way to validate innovation.
Arduino Nicla Product Series
A new system is about to be widely used, and the Internet of Things is another potential application that is becoming widespread.
We live in a tumultuous period of evolution. For example, let’s think about the wireless earbuds we use for our phones: they are a concentrated embodiment of artificial intelligence, which among other things can enhance audio quality and extend battery life. MEMS accelerometers are also a very interesting world, and we are confident about this. Arduino has partnered with Bosch and Bosch Sensortech to jointly develop systems in this field. Today, accelerometers may be created for different applications, and they can also be used for predictive maintenance with excellent performance. New sensors can detect particles in the air, which is not only very important in terms of pollution but also has very significant applications in fire detection and management. This innovation allows even “older” and currently quite cheap technologies to gain new opportunities. We take radar as an example. Today, advanced radar using ultra-wideband is installed in mobile phones: these technologies are becoming common, low-cost, and inspire us to continue developing systems. It is no coincidence that in the coming months, we will announce new products utilizing ultra-wideband technology, which will allow access to this technology in an open-source manner, not only as access to hardware but also as a simplification of object programming.
The adoption of Zephyr is also part of the viewpoint of technological democratization and ease of use; could you elaborate on that?
Zephyr is a new operating system that is very important to us. For various reasons, the operating system standardization process that has occurred in the microprocessor industry has not happened in the microcontroller industry. We have been using the next generation of these components for several years, which has actually changed the scheme: while the previous Arduino Core (the foundation of Arduino) was adjusted for specific processors and ran directly on them without an operating system, in the new products (starting from Arm’s Cortex M4), we have adopted a layered approach with an integrated operating system. The evolution of this process led us to choose Zephyr, a real-time operating system that is conceptually inspired by Linux, with absolute software portability. A revolution also stems from the experience of the COVID-19 pandemic, during which the difficulty of adapting new applications to different embedded systems based on microcontrollers due to component shortages became evident.
Today, with Zephyr and the new system layering technology, it is easy to adapt applications on different chips and different architectures. Furthermore, Arduino on Zephyr adds extra portability, allowing you to write code in Arduino language, and everything will still work optimally just by changing the architecture. Importantly, it is an open-source project, so there are many contributions from the GitHub community.
Arduino firmly believes in Zephyr, and we immediately contributed to this true revolution by joining that alliance. It is no coincidence that we will soon be able to see the first Arduino products equipped with the new operating system.
You are also one of the staunch supporters of Matter within Arduino. What is Matter?
Let’s start with a concrete example that almost everyone has encountered: a few years ago, I renovated my house, and to achieve some home automation, I bought some IoT objects. To set them up, I basically had to download different applications for each manufacturer’s brand. Some of these things could work with virtual assistants like Alexa, some with Google Home, and some could work with neither. Some required you to download an unknown app, and the ultimate destination of the data collected in your home was questionable. In short, anyone who has dealt with such objects has encountered huge interoperability issues.
Matter is an extremely complex protocol, but its success lies in the fact that it (or more precisely, the Connectivity Standards Alliance behind Matter) has successfully brought together participants of all sizes. It essentially contains a description system for every IoT object (such as smart light bulbs of any brand or model) and allows for its configuration, regardless of specifications (via radio, cable, Wi-Fi, Bluetooth connections, through mesh systems, etc.).
As an extremely complex system, I posed the challenge to Arduino’s R&D department to make it accessible, creating an Arduino library to allow any user to build systems with “Matter-compatible” devices in an easy way.
We made efforts in two areas: we built a new hardware board, Nano Matter, with wireless and Bluetooth connectivity, and an extremely easy-to-use library. With just a few lines of Arduino code, you can register a device (even using QR), view it on the dashboard, and configure the device on various systems like Alexa, Google, Apple, etc.
We are working with Silicon Labs to complete all this, as Silicon Labs is a leader in this technology, playing a very important role in supporting design and standardization, as well as in the relevance of the Matter underlying alliance.
We have released the “community preview” version of Nano Matter on the market, and we will release the final version in the coming weeks. We hope more and more people will get involved in building IoT devices to democratize this technology.
How do you think users (not just experts) will use technology, especially Arduino systems, in the near future?
There is a small premise, which is one of our corporate missions, and also one of my personal missions: I want people, especially young people, to no longer be “bystanders” of technology, but to become users and creators of technology. I see kids playing video games, and I think these kids may design new phones that others will play with in 15 years. Everything we do will have the necessary educational impact on society. We cannot let a small group of people control all technological innovation in the world while billions of others merely “suffer.” Therefore, we aim to unleash the creativity and enthusiasm of everyone through Arduino.
In terms of popularity, community, and most importantly, relationships with technology producers, we are at an advantage, positioning ourselves somewhat like the “Robin Hood” of technological innovation. We go to talk to high-level figures, vice presidents of semiconductor companies around the world; we ask them what they will do in the next three years and if they can let us know early what they are doing so that we can study, understand, and democratize it. It is for this purpose that we ask for significant discounts on technology costs to deliver innovation to everyone. I must say that we have a strong appeal to the top figures of these companies because everyone wants to expand their reach through their technology. Therefore, what we are doing, and will continue to do, is to try to understand two things: on one hand, the new problems that can be solved; on the other hand, how existing technologies can provide new solutions to existing or still potential problems. For example, LiDAR in mobile phones.
We built a robot called Alvik, which, unlike other robots equipped with ultrasonic sensors, has an integrated LiDAR sensor. Through Alvik, young people learn to use LiDAR, and maybe tomorrow they will invent a 3D scanner based on that technology. Ultimately, our goal is to bring technologies that may be used in a certain field to the mass market, where many creators will be able to use them to create new applications and new solutions. This is our vision. Obviously, we know that the user base is also constantly evolving. In the past, users were more patient; they needed to understand electronics and soldering wires in detail. Arduino has always been project-based, so it is not about electronics for the sake of electronics. With Arduino, the first thing is to solve a problem, and then you understand what you did. Today, there is a lot of research on instant gratification, especially in the new generation. Young people’s time is becoming increasingly scarce, and they are burdened with studies, homework, extracurricular activities, etc. They do not have much time, and their parents also do not have the time or patience to accompany them in learning or playing. Therefore, based on this evolution, we have created plug-and-play kits that accurately represent the concept of “I solve a problem, and then I can understand it in more detail,” but using modern means, just as today’s children want it. Another thing we understand is the importance of connectivity, and thus the possibility of using mobile devices (with many sensors on board) to control applications and generate data. We have made great efforts in this regard: for example, Arduino Cloud allows you to easily build connected devices in “Arduino style” within ten minutes. My request to anyone is to enable everyone to develop IoT applications faster than ever with Arduino.
Using the Arduino Plug and Make Kit to create a video game controller
All you need is a connection card, and with the plug-and-play kit, you can conduct many experiments. We are working hard and will see the capabilities of Arduino Cloud expand, especially in terms of IoT and security features, as this is an extremely important aspect for us. Since I joined Arduino, we have equipped every board with encryption chips. Initially, they mocked me a bit for wanting to add two euros of cost to a 20-euro product. But security is crucial, and once compromised, it cannot be undone. Furthermore, it also secures applications and prevents hackers from accessing your home devices and doing as they please. We are introducing new programming languages, including those most commonly taught in universities, such as Python and MicroPython. We have many projects planned for the future, which will become products from the end of this year to 2025, also based on MicroPython, providing increasingly different possibilities.
Considering the development of artificial intelligence, wanting to look further into the future, perhaps the next step may be programming with natural language. This method can already be well applied to Arduino language, as generative models can learn from a wealth of online resources: from community-shared projects to the specifications of our hardware products, everything is widely available. Therefore, artificial intelligence can understand how code works and what it can do, and then provide useful solutions at the first moment.
Compiled from MIT Technology Review
Welcome to set us as a “star,” so you can receive push notifications in the first place.
Scan to follow: Automotive Development Circle, reply “Auto”
Get free Autosar introductory and practical material package!
Scan to add the assistant and reply “Join Group”
Face-to-face communication with electronic engineers to exchange experiences
