As its website explains: “Coral works with the best of Google’s ML tools, including TensorFlow and Cloud… We’re exploring new ways to implement AI by bringing it local, making it fast, reliable, and respectful of user data.”
The new beta dev board boasts an Edge TPU platform and a custom-made ASIC to run meta language algorithms “at the edge.” Basically, Google built what looks like a Raspberry Pi with machine learning.
The American giant explains that this hardware is perfect for prototyping new projects and internet-of-things (IOT) devices, which require fast on-device inferencing for machine learning models.
Google adds that it’s also possible to scale from prototype to production using just the 40 mm × 48 mm system-on-module (SOM) board combined with custom PCB hardware using board-to-board connectors.
This new hardware was launched before TensorFlow Dev Summit, an event that brings together machine learning users for technical conversations with the TensorFlow community.
A Few Technical Details…
The dev board is a single-board computer which supports Mendel Linux and TensorFlow Lite. Languages include Python and C++ is coming soon. Features include:
- Edge TPU Module (SOM)
- NXP i.MX 8M SOC (Quad-core Cortex-A53, plus Cortex-M4F)
- Google Edge TPU ML accelerator coprocessor
- Cryptographic coprocessor
- Wi-Fi 2×2 MIMO (802.11b/g/n/ac 2.4/5GHz)
- Bluetooth 4.1
- Flash is provided with 8GB eMMC
- 1GB LPDDR4
- Connectivity is through USB
- Integrated GC7000 Lite Graphics
- MicroSD card slot
- Gigabit Ethernet port
- 40-pin GPIO expansion header
If you’re interested in purchasing the Google Coral Dev Board, you can expect to pay $149.99. You can call Mouser to purchase this product.
As well as the Dev Board, Google has further products available including a USB accelerator, camera and PCI-E device which promises easy integration of Edge TPU into existing systems.