With the new VENTUNO Q board, Arduino hopes to simplify robotics development

Arduino Srl, a subsidiary of Qualcomm Technologies Inc., recently opened pre-orders for VENTUNO Q, an edge AI platform based on the Qualcomm Dragonwing IQ8 series. Arduino made the board for developers, prioritizing familiarity and ease of use with the right software, accessories and extensions.

While this is Arduino’s first intentional step into robotics, Marcello Majonchi, Arduino’s chief product officer at Qualcomm, said the company has more than 20 years of experience working with robotics developers.

“This is the first intentional step for a product intended to be used as the single point of control of a modern robotic platform,” he said The robot report. “At the same time, Arduino is widely used in the field of robotics and there are thousands of different types of robots that, over the last 20 years, have been developed by the community and enhanced in some functionality or in their entirety with an Arduino product.”

Qualcomm integrates Arduino technology

Last year, Arduino was acquired by Qualcomm, which quickly gave the Monza, Italy-based company access to more resources and a new customer base. This, combined with recent advances in generative artificial intelligence and processor size, made the new card seem like a natural evolution, Majonchi said.

“We have this amazing computing power. We have this ecosystem that’s just waiting for something that offers an easier way for people to translate their vision into reality,” he noted. “What is the gap that we need to fill from a hardware perspective and from a software tools and frameworks perspective so that you don’t need five different PhDs to figure out how to build an AMR?”

Technical details of the VENTUNO Q

TWENTY-ONE Q has a dual-brain architecture. It pairs a Dragonwing IQ-8275 processor, offering up to 40 dense TOPS of AI performance, with 16GB LPDDR5 memory, 64GB eMMC, and expandable M.2 NVMe storage along with a dedicated STM32H5 real-time microcontroller. Arduino said this enables systems that don’t just interpret sensor data, but also respond physically, immediately and offline.

The card comes preloaded with Ubuntu, distributed by Canonical. Arduino App Lab offers a quick path to a working prototype. Developers can also choose VS Code, native Linux tools, or any existing workflow. On the MCU side, the Arduino Core on Zephyr RTOS is designed to provide deterministic, real-time control of motors, CAN-FD, PWM, and high-speed GPIOs.

When building the board, Arduino said it wanted to ensure that much of the things customers have come to expect from any of its other boards carry over to the VENTUNO Q as well. That’s because there is a large ecosystem of accessories and add-ons that are compatible with a standard Arduino board. Replicating that system gives customers an option they are familiar with, making it easier to assemble different components, the company said.

“It’s all about accessibility, in the sense of breaking down the barrier of entry to this technology and demystifying it on both the software side and the hardware side, so that almost anyone who has an idea has a path to bring that idea to life,” Majonchi said.

Edge computing will be critical for real-world deployments

A key aspect of the VENTUNO Q card is its ability to provide edge computing for robots.

“One of the obstacles that prevented the robotics industry from entering the age of artificial intelligence was the fact that models, until just a couple of years ago, were stuck in the cloud,” Majonchi said. “If you want to make a robot work, you need more than just near real-time response. You need real-time response, and you need to have, ideally in the same platform, the sensing, reasoning, and actuation to orchestrate all the behaviors.”

“[Edge computing] it’s the only way to really meet many of the constraints of real-world deployment of these robots that might end up in warehouses or offshore manufacturing sites, where you can’t even expect a reliable high-speed connection to the Internet,” he continued.

Technology has advanced to make this level of cutting-edge processing possible, Majonchi noted. Combined with the high power-to-compute performance ratio, this created a good starting point for Arduino, he said.

Arduino App Labs allow customers to build robotic applications one brick at a time

Through the Arduino App Lab, developers can immediately run a curated library of NPU-optimized models. These include large language models (LLMs), visual language models (VLMs), speech recognition, and object and gesture detection.

Developers can also bring their own model in GGUF format from Hugging Face or train and quantize custom models via the built-in Edge Impulse Studio. LLM and VLM can also be deployed in just a few lines of code using the Qualcomm Gen AI Developer Tool (GenieX), Qualcomm’s open source on-device Gen AI runtime.

Majonchi compared App Lab to building with Legos. “You just assemble those bricks and recreate the connection code, which is a lot easier than what you would do on a normal platform out there,” he said.

The company designed and validated the platform for commercial deployment in robotics, cutting-edge artificial intelligence, smart cities, AI vision systems and education.

“We created a development environment that didn’t invent new technologies, but simply abstracted existing technologies, containers, ROS, and AI models into one place where people would find inspiration and a framework that was easy to approach and yet non-binding,” Majonchi said.

Works with Arduino hopes to make commercial implementations possible

Once a VENTUNO Q design is ready for commercial deployment, the same Dragonwing IQ8 processor will soon be available from partners in production-grade SOM form factors, preserving the software stack, AI models and application logic created during prototyping.

SECO, a leading provider of industrial edge computing and artificial intelligence solutions, and Toradex, a specialist in industrial-grade SOM, are among the first to support this journey as part of the “Works with Arduino” program.

“Arduino is the ultimate prototyping platform,” Majonchi said. “Enthusiasts, teachers and professionals have standardized how they prototype across many different industries, from automotive to space.”

“At the same time, there are some specifics of a production setup that are a little different than a prototyping setup,” he added. “There’s an optimization aspect. There’s long-term availability of components and so on.”

“So because we’re now part of the Qualcomm family, we’ve created a program that we announced in conjunction with the release of TWENTY-ONE Q that is trying to really support end customers who want to have the freedom and ease of prototyping, but then want a reliable path to get to production in the most optimized way,” Majonchi said.

Beyond prototyping, VENTUNO Q is designed for commercial deployment, powering everything from autonomous mobile robots and predictive maintenance to real-time quality inspection and hands-on STEM learning.

“Works with Arduino is really aimed at OEMs who can basically take the same Qualcomm chipset that’s in our development boards and build any type of solution, from systems and modules to industrial computers and industrial gateways that can be in any type of robot, and be certified against the Arduino platform,” Majonchi said.

Arduino continues to grow with Qualcomm

Looking to the future, Majonchi said Arduino is interested in continuing to learn and grow in the robotics industry. The parent company, Qualcomm, will play a big role in this.

“Our expertise is in providing this level of abstraction for people to benefit from technology. What we have historically lacked is access to cutting-edge technologies that we could adopt and bring to the masses. Together with Qualcomm, we now have both sides of the equation, so we can solve the problem,” Majonchi said.

Qualcomm is also showing more interest in robotics. The company said today that it is acquiring PickNik Inc., the publisher of the open source MoveIt framework. The companies plan to integrate MoveIt with Arduino VENTUNO Q and Qualcomm Dragonfly to give developers a simpler path from AI models to real-time planning, manipulation and control.