Google deploys machine learning to reduce cost of edge computing for manufacturers

Internet giant Google is using machine learning to reduce the cost of edge computing for manufacturing and automotive companies.

Injong Rhee, vice-president of IoT at Google Cloud, said that many benefits can be derived from intelligent real-time decision-making such as manufacturing companies detecting anomalies in high-velocity assembly lines. But these real-time IoT (Internet of Things) systems pose challenges due to costs, form-factor limitations, latency, power consumption, and more, Rhee added.

In order to address these, Rhee said that the internet giant was looking at bringing machine learning to edge computing by providing a new hardware chip called Edge TPU and a software stack called Cloud IoT Edge that extends Google Cloud’s artificial intelligence abilities to IoT gateways and connected devices.

"This lets you build and train machine learning models in the cloud, then run those models on the Cloud IoT Edge device through the power of the Edge TPU hardware accelerator," Rhee explained.

According to the company, the hardware -- Edge TPU -- is a chip designed to run TensorFlow Lite Machine Learning models at the edge of connected devices.

"When designing Edge TPU, we were hyper-focused on optimising for 'performance per watt' and 'performance per dollar' within a small footprint. Edge TPUs are designed to complement our Cloud TPU offering, so you can accelerate machine learning training in the cloud, then have lightning-fast machine learning inference at the edge. Your sensors become more than data collectors — they make local, real-time, intelligent decisions," Rhee explained.

This means that companies working on object recognition such as autonomous vehicle firms can install GPUs/CPUs to make decisions faster and avoid untoward situation.

The software stack, Cloud IoT Edge, helps execute faster edge processing, Google said. The firm added that it can run on Android Things or Linux OS-based devices, and its key components include a runtime for gateway class devices, with at least one CPU, to locally store, translate, process, and derive intelligence from data at the edge. Cloud IoT Edge seamlessly operates with the rest of Cloud IoT platform and the Edge IoT Core runtime, which securely connects edge devices to the cloud, enabling software and firmware updates and managing the exchange of data with Cloud IoT Core.

Rhee also said that since the firm wanted developers to start testing out the Edge TPU chip, it has built a development kit that includes a system on module (SOM) that combines Google’s Edge TPU, an NXP CPU, Wi-Fi, and Microchip’s secure element in a compact form factor. It will be available to developers this October, the IoT head said, adding that the company was also working with ecosystem partners.

"Semiconductor partners will create the SOM with the Edge TPU chip inside. Device makers will make industrial IoT gateways—like the kind used in factories, locomotives, oil rigs, and more—that include the SOM and Edge TPU. Partners include semiconductor companies such as NXP, ARM; gateway companies such as Accton, Harting, Hitachi Vantara, Nexcom and Nokia; and edge computing partners such as Adlink Technology, Kelvin, Olea Edge Analytics, Smart Catch and Trax," Rhee said.