Buzzword buzzkill: Cloud, AI, IoT and edge work in a real product
SAP teaches factory equipment to pre-analyse for costly faults, call cloud for help
Artificial intelligence, the internet of things and edge computing are 2017's inescapable buzzwords, and cloud probably has that role for the entire decade. So imagine The Register's surprise when we learned all three are working together in a product you can put to work today.
That product is SAP's Leonardo IoT Edge that, as described to El Reg by the company's senior veep for IoT Smart Connected Business Elvira Wallis, helps manufacturers detect potential anomalies before they turn into costly messes.
“Today you have a production run and the quality assurance takes place after the run is done,” Wallis explained, offering up humidity's impact on paint jobs in a car factories as an example of tiny variables that can have big impacts on the final product, sometimes years after manufacturing. SAP's betting that manufacturers would rather know about the potential for a problem before a production run is finished, rather than after a batch has been botched.
Enter edge computing, in this case a Dell gateway – aka a ruggedised PC/server – hooked up to industrial equipment and running SAP Edge Services. If an organisation's industrial equipment has an API and spits out well-formed data, SAP can be taught how to parse it and therefore how to spot anomalies that could impact product quality. If output is more exotic, SAP and/or its partners have a deeper services engagement to undertake, but the result is the same – the gateway gets the data it needs to monitor for quality.
The gateway will be loaded with what Wallis called a “golden batch formula” for equipment's ideal performance. If the machine varies from that formula, warnings are sent to relevant folks before product quality drops.
Data fed to the gateway can be pre-processed there and/or allowed to escape to SAP's Leonardo cloud so that weightier processing power and a dash of AI can be brought to bear on data the gateway thinks needs more immediate attention – and the time and expense required to take data all the way to the cloud and analyse it there.
The gateway can also decide to send data upstream for deeper analysis at a convenient time, rather than in a screaming rush when an anomaly appears. After such leisurely analysis, the software will push revisions to the golden batch formula down the gateway so that monitoring on the edge uses more accurate parameters.
Wallis said an early user of the product makes carpets used in big venues and spotted a tiny mistake that would have made different batches look subtly different under the same lighting conditions. While visitors might not have noticed, venue owners probably would. A fix ensured consistency.
So there we have it: cloud, AI, IoT and edge computing, in harness, in a real application you can buy now. We'd better go looking for some new buzzwords about which to be suspicious! ®