Meet the new Grove AI HAT: a $25 Raspberry Pi board for Edge Computing 9th May 2019 Edge computing is an innovation already in consumer homes, without us even knowing it. Now it’s even more affordable, thanks to Seeed Studio’s $25 Grove AI HAT.Like the movies warned us, the AI revolution hasn’t announced itself with fanfare so much as silently infiltrating our lives. With the advent of the new Grove AI HAT, edge computing workloads and Raspberry Pi projects can be even easier and more affordable than ever. The Grove AI HAT is a board that comes with RISC-V capabilities and can be used for AI vision and voice recognition projects, like the ones that the likes of Tesla, Apple and Amazon are producing for the masses. This means that the board will process data and machine learning on-device, without relaying information back to the cloud. In the world of IoT, think of the Amazon Echo as a shining example for the capabilities of edge computing: the fact that commands are processed by Alexa on the device itself, makes for a more efficient experience with no latency, but it also means increased security. This is exactly what the new Grove AI HAT offers for using Raspberry Pi, at a low cost. The Grove AI HAT comes with four Pi mounting holes and integrates the full ArduinoCore-API interface to support Arduino IDE, Linux, Windows, Mac OS X, and other development environments. Seeed Studio promises to bring their hundreds of grove senors and grove actuators to AI applications with the new board, and for AI voice recognition applications, there’s a microphone. What is RISC-V? The Grove AI HAT board comes with RISC-V capabilities too – pronounced “Risk Five” – which is an instruction set architecture (ISA). It’s RISC-V that describes how your software reacts to a processor, but the truly exciting thing about RISC-V is that it’s open source. Anyone can build a processor to support RISC-V. Additionally, anyone can create their own implementation of RISC-V processors, meaning that a processor can implement custom extensions, whatever the given application. Foundation members of the RISC-V foundation oversee the standard, with a number of free open-source implementations already out there. All this simply means a greater range of options for customers. “Anyone can create their own implementation of RISC-V processors, meaning that a processor can implement custom extensions” The RISC-V Foundation has along with the Linux Foundation, partnered with Western Digital and Google to promote the development of RISC-V implementations. The project, named the CHIPS Alliance, has seen Western Digital commit to “shipping two billion RISC-V cores annually,” once they have transitioned their designs to RISC-V. The board breakdown The Grove AI HAT is built around a Sipeed MAIX M1 AI module. This is a powerful RISC-V module that contains a Kendryte K210 processor; it can work independently for edge computing applications and assist Raspberry Pi to run AI. The MAIX module is according to Sipeed, “a master controller, not an accelerator,” and can be used for predictive maintenance, machine vision, robotics and voice recognition. The K210 CPU at the heart of the MAIX M1 has a neural network processor that is capable of object recognition. The Grove AI HAT also comes with full support for ArduinoCore-API. This means that the HAT will utilise Grove Arduino Libraries, which can allow users to connect to sensors and modules. “For AI voice recognition applications, we added a high-quality microphone… and for robot or motion applications, there is an onboard 3-axis accelerometers sensor, which is more accurate and easy to use compared to external sensors,” Seeed Studio claim about the new Grove AI HAT for Edge Computing. “We hope this board may help you with your edge computing, AI vision, voice recognition, and other AI project,” the studio added. Seeed Studio is currently offering the Grove AI Hat for pre-order for US$24.50. It is set to ship worldwide from June 15th, with prices on the board rising to $28.90 after that date. Photo from Seeed Studio Article written by:Mark WhiteMark is a writer/editor who has written online and in print.