January 21, 2020
AI can enable many important tasks from manufacturing to medicine, but only if the applications are speedy and secure. Communication via the cloud adds latency and risks privacy, which is why Google worked on a solution — dubbed Coral — that avoids centralized data centers. Coral product manager Vikram Tank described Coral as a “platform of [Google] hardware and software components … that help you build devices with local AI — providing hardware acceleration for neural networks … right on the edge device.”
The Verge reports that Coral, which just exited beta last October, is “part of a fast-growing AI sector” that analysts predict will sell more than “750 million edge AI chips and computers … in 2020, rising to 1.5 billion by 2024.” Most, but not all of them, are destined for consumer devices such as smartphones. Many others, however, will be picked up by enterprise customers in various industries.
Coral offers both “accelerators and dev boards meant for prototyping new ideas, and modules that are destined to power the AI brains of production devices like smart cameras and sensors.” The hardware for both is Google’s Edge TPU, “an ASIC chip optimized to run lightweight machine learning algorithms.”
Also provided is a guide on creating “fun projects,” such as an AI marshmallow-sorting machine and a smart bird feeder. According to Tank, “the long-term focus … is on enterprise customers in industries like the automotive world and healthcare,” including autonomous vehicles. Tank pointed out that any latency would be problematic for “a car moving at 65 mph [that] would traverse almost 10 feet in 100 milliseconds.” In healthcare, the solution also enables “real time analysis of ultrasound images using image recognition,” whereas sending those images to the cloud “creates a potential weak link for hackers to target.”
Coral, which has its roots in Google’s AIY learning kits launched in 2017, has competition that runs “the gamut from startups like Seattle-based Xnor, which makes AI cameras efficient enough to run on solar power, to powerful incumbents like Intel, which unveiled one of the first USB accelerators for enterprise in 2017 and paid $2 billion last December for the chipmaker Habana Labs to improve its edge AI offerings.”
The Coral team said “it differentiates itself by tightly integrating its hardware with Google’s ecosystem of AI services … [such as] a library of AI models specifically compiled for its hardware, as well as AI services on Google Cloud that integrate directly with individual Coral modules like its environment sensors.” One potentially limiting factor is that Coral’s “Edge TPU-powered hardware only works with Google’s machine learning framework, TensorFlow.”
A spokesperson for AI edge firm Kneron noted that, “Coral products process specifically for their platform [while] our products support all the major AI frameworks and models in the market.” Google has been mum on Coral sales, leaving the number of enterprise purchases unknown.