New hardware complements Edge TPU chip for easy deployment of AI solutions
AI just got a lot more accessible with the official launch of Google’s Linux-powered Coral Dev Board and USB accelerator, announced last summer at the Google Next conference and presented at the TensorFlow Dev Summit this week. The new hardware will work together with Google’s Edge Tensor Processing Unit (TPU) chip, a tiny, purpose-built ASIC designed to deploy ML inferencing at the edge.
Trained in the cloud, Edge TPU offers high-accuracy AI without the bulky power footprint, meeting current demands for privacy, low latency and bandwidth constraints. It can be used for a wide range of applications such as predictive maintenance, machine vision, robotics and voice recognition in various fields.
Linux-powered Coral Dev Board and USB Accelerator enhance systems with Edge TPU
The Coral Dev Board is a single-board-computer featuring a removable SOM with an embedded multimedia card, system-on-chip, wireless radios and an Edge TPU coprocessor.
The Coral Dev Board is ideal for prototyping new projects that demand fast on-device inferencing for machine learning models in IoT devices and other embedded systems. Users can easily scale to production by combining custom PCB hardware and the SOM board using board-to-board connectors.
Connected via USB socket, the USB Accelerator adds an Edge TPU coprocessor to Linux-based systems such as Raspberry Pi, enabling accelerated ML inferencing. The USB accelerator is compatible with TensorFlow Lite models.
AI startups will benefit from Google's accessible AI tools
One of the main setbacks for budding AI startups is the enormous power requirements needed to train machine learning models. Thanks to its low power and computational requirements, virtually anybody can take advantage of the Google Edge TPU and integrate it with their existing systems to boost their machine learning capacities. This is a game-changer for startups, who often have limited budgets and/or facilities.
Google now offers pretty much the whole gamut to product developers in one efficient package, from the cloud TPU to the inference and machine training hardware. At $150 for the Coral Dev Board and $75 for the USB Accelerator, both of the latest hardware products come at a reasonable price point.
The new Coral Dev Board and USB Accelerator are sure to become invaluable tools in empowering product developers and facilitating the creation of prototype IoT devices by startups around the world.