Artificial Intelligence at the Edge is revolutionising the way we develop smart solutions, enabling data to be processed directly where it is generated. This reduces latency, improves privacy, optimises energy consumption and makes it possible to deploy AI applications on increasingly smaller and more efficient devices.
With the aim of facilitating access to these technologies, the dAIEDGE-VLab brings together a catalogue of hardware platforms that allows researchers, developers, businesses and students to experiment with different Edge AI architectures without needing to physically possess the devices. From Linux systems for prototyping to platforms specialising in artificial intelligence or microcontrollers for low-power applications, the virtual laboratory offers a versatile environment for validating solutions tailored to different use cases.
Among the available platforms, the Linux Edge Boards stand out, represented by the Raspberry Pi 5 and Raspberry Pi 4B – two of the most widely used development boards for edge computing, IoT and embedded applications. Equipped with quad-core ARM processors, they provide a complete Linux environment and an excellent combination of performance, connectivity and ease of development, making them an ideal choice for rapid prototyping and the deployment of smart applications.
For projects that require running artificial intelligence models directly on the device, the laboratory incorporates several AI Edge Platforms. The Jetson Orin Nano offers high performance for computer vision, robotics and deep learning inference applications in a compact form factor. The Qualcomm RB3 Gen 2 is designed for autonomous systems and intelligent robotics, incorporating hardware acceleration for AI with low latency. The Kendryte K210, based on RISC-V architecture, incorporates an AI accelerator that enables computer vision and machine learning tasks to be carried out with very low power consumption. Rounding off this category is the SynSense Speck™ Dev Kit, a neuromorphic computing platform that combines an event-driven vision sensor with a spiking neural network processor, delivering real-time AI inference with exceptionally low power consumption.
The laboratory also offers users a selection of embedded controllers, designed for applications where real-time operation, energy efficiency and reliability are essential. The STM32MP257 combines ARM Cortex-A35 and Cortex-M33 cores with a Neural Processing Unit (NPU), enabling Linux, control tasks and AI inference to be run on a single platform.
For its part, the LPC55S69 features two ARM Cortex-M33 cores, advanced security features via TrustZone and DSP support, whilst the STM32L4R9 stands out for its ultra-low power consumption and its ability to develop battery-powered embedded and IoT applications.
The combination of all these platforms makes our dAIEDGE-VLab a unique environment for exploring the potential of Artificial Intelligence at the edge. Users can compare architectures, evaluate how the same application performs on different devices, and select the most suitable hardware based on performance requirements, energy consumption or processing capacity. From general-purpose Linux systems to AI accelerators and emerging technologies such as neuromorphic computing, the laboratory offers an infrastructure ready to drive the next generation of Edge AI solutions and bridge the gap between research and real-world applications.