Arduino has opened pre-orders for VENTUNO Q, designed to combine high-performance local AI with deterministic control for robotics and physical AI applications.
Edge AI is moving beyond systems that only classify images, recognise speech or detect anomalies. The next step is physical AI: machines that interpret their surroundings, make decisions and act locally. This shifts developer expectations of what they need from an embedded platform. AI inference is only part of the solution when a robot must also steer, grip or stop.
VENTUNO Q aims to bring those functions together on one board. Its dual-brain architecture is structured to pair a Qualcomm Dragonwing IQ8 processor with an STM32H5 real-time microcontroller from STMicroelectronics.
The processor is designed to handle demanding AI workloads, while the microcontroller is constructed to manage deterministic control for motors, CAN-FD, PWM and other time-sensitive interfaces. Arduino describes this as bringing perception, decision and action into a single platform. The approach targets applications where intelligence must translate directly into physical behaviour.
From vision to movement
The practical value becomes clearer in robotics. An autonomous mobile robot can use Linux and ROS 2 for navigation and perception, while at the same time, the microcontroller can handle real-time motor control and other physical responses. Vision-guided manipulators can similarly connect object recognition with precise movement.
The same architecture extends beyond robotics. Arduino identifies automated quality inspection, predictive maintenance and process automation among the platform’s potential industrial applications.
Local processing also supports systems where continuously sending data to the cloud is undesirable. Traffic monitoring and smart environments can instead process information on-device, such as for offline voice detection and multimodal applications.
Developers can start by choosing a ready-to-use, NPU-optimised model from the growing library made available within Arduino App Lab, powered by Qualcomm AI Hub, or train custom models through the integrated Edge Impulse platform. Arduino App Lab is designed for seamless versatility because it brings Python, Arduino sketches, and AI models into one development environment, and also allows for the use of standard Linux tools, including VS Code and Docker. VENTUNO Q ships with pre-loaded Ubuntu and is bundled with an Ubuntu Pro license.
A route beyond the prototype
Through the Works with Arduino programme, certified third-party system-on-modules can run compatible Arduino App Lab applications without code changes. SECO and Toradex are among the first partners offering production-grade modules based on the same Dragonwing IQ8 architecture.
This creates a route towards production hardware while carrying forward software, models and application logic developed during prototyping. The board is also designed to support Arduino UNO shields and carriers, Arduino Modulino nodes and Raspberry Pi HATs. Existing sensors and actuators can therefore remain part of the development environment.
VENTUNO Q is engineered to provide up to 40 dense TOPS of AI acceleration, alongside 16 GB LPDDR5 RAM and 64 GB eMMC storage. These resources support applications combining local perception, reasoning and physical control.
Overall, the platform extends Arduino’s familiar prototyping model into physical AI, where inference is only one part of the system. The important goal for Arduino is advancing not simply what a machine can recognise, but what it can do with that information.
VENTUNO Q is available for pre-order through Arduino and authorised distributors, including Farnell, Mouser, DigiKey and RS.
For more electronics updates please visit: https://efemag.co.uk/category/news
