Google Local AI - the Mini Edge TPU Dev Board launched at the new year!

DIY MAKER 於 20/07/2020 發表 收藏文章
Google launched a series of hardware such as development boards based on Edge TPU chips in March 2019, as well as a localized AI platform - Coral. However, just at the beginning of this year (January 2), Google unexpectedly posted on its developer blog that it would launch a Coral Dev Board Mini based on the Edge TPU chip in the first half of 2020.
What is Edge TPU chip
Edge TPU is an ASIC chip designed for running TensorFlow Lite ML models on the edge. Edge TPU can be used in more and more industrial use scenarios, such as predictive maintenance, anomaly detection, machine vision, robotics, speech recognition, etc. It can be applied to various fields such as manufacturing, local deployment, healthcare, retail, smart space, and transportation. Its small size and low energy consumption, but excellent performance, can deploy high-precision AI on the edge. Edge TPU complements CPU, GPU, FPGA, and other ASIC solutions that run AI at the edge.
The following is the full text of the Coral Dev Board Mini from the Google Developer Blog:
More and more industries are beginning to realize the value of local AI, where the speed of local inference can save a lot of bandwidth and cloud computing costs, while keeping data locally can protect user privacy.
Last year, we launched Coral, our hardware components and software tool platform, which can easily prototype and extend local AI products. Our product portfolio includes Coral Dev Board, USB Accelerator and PCIe Accelerators, which are now available in 36 countries.
Since its release, we have been pleased with Coral-based applications that cover many industries from healthcare to agriculture to smart cities. By 2020, we are happy to announce the new features of the Coral platform, which will further expand scalability.
The first is the Coral Accelerator Module, which is an easy-to-integrate multi-chip package that encapsulates the Edge TPU ASIC. The module provides both PCIe and USB interfaces, which can be easily integrated into custom PCB designs. We have been working closely with Murata Manufacturing Co. to produce this module. The Coral accelerator module is expected to be launched in the first half of 2020.
Next, we announced the launch of the Coral Dev Board Mini, which can replace the Coral Dev Board to provide a smaller form factor, lower power consumption and lower cost. Mini combines the new Coral Accelerator module with the MediaTek 8167s SoC to create a board that excels in 720P video encoding/decoding and computer vision use cases. The motherboard will be on display during the International Consumer Electronics Show in 2020. Coral Dev Board Mini will be available in the first half of 2020.
We also provide a new version for Coral System-on-Module, in addition to the original 1GB LPDDR4 configuration, now also offers 2GB and 4GB LPDDR4 RAM. We will work with the NXP team at the 2020 International Consumer Electronics Show (CES 2020) to show how to use SoM in smart city, manufacturing and healthcare applications, as well as some new SoCs and MCUs that we have worked with the NXP team explore.
Finally, Asustek chose Coral SOM as the basis for its Tinker Edge T product, which is a manufacturer-friendly single-board computer with rich I/O interfaces, multiple camera connectors, programmable LEDs and color-coded GPIO headers .
AI hardware and software supporting Google Edge TPU edge artificial intelligence computing chip
In addition to the big manufacturers NXP and ASUS that were mentioned in the above question, the Model Play and Tiorb AIX (Tiorb AIX) launched by Gravitylink can also perfectly support Edge TPU. AIX is an artificial intelligence hardware that integrates two core functions of computer vision and intelligent voice interaction. Model Play is an AI model resource platform for global developers. It has built-in diversified AI models, combined with Tiorb AIX, and is based on Google open source Neural network architecture and algorithm, build autonomous migration learning function, no need to write code, AI model training can be completed by selecting pictures, defining models and category names.

留言


請按此登錄後留言。未成為會員? 立即註冊
    快捷鍵:←
    快捷鍵:→