Breaking Through the Last Bottleneck in Manufacturing’s Digital Transformation
Sight Machine Inc. announced it is collaborating with NVIDIA Corp. to apply machine learning to turn the chaos of factory data into insights for improving production.
The collaboration connects Sight Machine’s manufacturing data foundation with NVIDIA’s AI platform to break through the last bottleneck in the digital transformation of manufacturing – preparing raw factory data for analysis. Sight Machine’s manufacturing intelligence will guide NVIDIA machine learning software running on NVIDIA GPU hardware to process two or more orders of magnitude more data at the start of digital transformation projects.
Sight Machine’s manufacturing data foundation solves the central challenge in the digital transformation of manufacturing, turning raw plant data into a data foundation that captures manufacturing in digital form. Sight Machine then applies data analytics and AI/ML in real time to solve the core problems in manufacturing, which include improving performance measures like throughput, quality and downtime, and sustainability measures including energy use and scrap.
Read More:Â Using Marketing Automation To Optimize Your Sales Game
However, a bottleneck – common to all complex data environments but most difficult in manufacturing – remains near the front end of the process: data labeling, sometimes referred to as tag mapping. Before a data stream can be incorporated into a data model or data foundation, one must know what that particular type of data represents and where it came from. To apply AI at scale, enterprise manufacturers must understand the data they analyze.
A modern factory may generate streams of data from 100,000 or more point sources, such as individual sensors, and large enterprises must manage millions. Many companies have been collecting industrial data for years in a data lake or historian, assuming that once they get data aggregated they will be able to derive value from it. Few companies have managed to identify all those data points in a way that makes them useful.
Further complicating the picture: although historians and data lakes improve data accessibility, they often lose associated metadata from the PLC or other source devices. The result is a large number of data values that have been removed from their context, making it even more difficult to understand if or why the data is useful.
Sight Machine is aiming to break the data labeling bottleneck by linking its streaming data pipeline with the NVIDIA AI platform, running on Microsoft Azure infrastructure, to map data to assets at a global scale.
The collaboration with NVIDIA will enable Sight Machine to organize orders of magnitude more data than is currently feasible, at high speeds and without consuming excessive time from data scientists, controls engineers or other subject matter experts.
“This work addresses the last critical bottleneck in manufacturing transformation and will rapidly accelerate day-to-day use of AI in plants,” said Jon Sobel, CEO and Co-Founder of Sight Machine. “We’re taking our data discovery / introspection / analysis loop, with heuristics developed over a decade of mapping data, and turbocharging it with NVIDIA’s AI platform. This approach will prepare factory data for analysis at a previously unimaginable speed, bringing what currently takes two to three months of effort down to hours and days. This is all possible because of the decade of experience being brought to the problem through AI.”
“NVIDIA GPU acceleration of manufacturing will allow industrial companies to unlock trillions of dollars in value,” said Piyush Modi, global development leader and chief strategist for the industrial sector at NVIDIA. “Sight Machine’s adoption of the NVIDIA AI platform will help customers improve their understanding of data and ultimately speed up the digital transformation on their factory floors.”
Accelerating data labeling will enable Sight Machine to quickly onboard large enterprises with massive data lakes. It will automate and accelerate work and lead to even faster time to value. While similar automated data mapping technology is being developed for specific data sources or well documented systems, Sight Machine is the first to use data introspection to automatically map tags to models for a wide variety of plant floor systems.
Sobel compared factory data to shale oil. “This data wasn’t created or managed for analysis,” Sobel said. “We’re retrofitting data that was generated for a different purpose. For a decade people have been saying data is the new oil, but really, it’s more like shale. You can’t just use it; you’ve first got to force the oil out of the rock.”
Read More: Everest Group Names Genpact A Leader In Banking Operations – Services PEAK Matrix® Assessment…
Automating Today’s Process
In Sight Machine’s client onboarding process, data engineers work with plant floor veterans to identify the settings that most impact production. In a typical factory, engineers and subject matter experts might have between 10 and 100 go-to data tags (e.g., machine settings) they typically work with when needing to make adjustments to fix a problem or tweak output.
Sight Machine starts with those tags and expands the amount of data to be analyzed by an order of magnitude, relying on software automation, analytics and its expertise in identifying the data types most strongly connected with manufacturing performance. The resulting data foundation then focuses on as many as several thousand critical tags. However, this process still only incorporates a fraction of an enterprise’s data.
The new initiative will automate the labeling of all data points, combining Sight Machine’s decade of expertise in digitizing manufacturing with NVIDIA’s AI platform and expertise in deep learning, and with Microsoft’s end-to-end streaming and AI solutions for manufacturing. NVIDIA GPUs, originally developed for high-speed graphics, are ideal for crunching large volumes of data efficiently.
Even manufacturers several years into digital transformation projects often rely on only a fraction of their data. “There’s a huge opportunity if we can help them make sense of all of their data,” said Kurt DeMaagd, Sight Machine Chief AI Officer and Co-Founder.
Sight Machine is working with one of the world’s largest agricultural products companies to digitize its factories. Running on NVIDIA’s AI platform, Sight Machine will be able to automate data labeling for literally millions of tags at a small fraction of the time and cost.
“This is a way of creating some order out of the data chaos, so that either subject matter experts can then manually work with that data, or companies can apply additional tiers of machine learning on top of that data,” DeMaagd said. “This is the foundation for highly-scalable, higher-level machine learning projects.”
Read More:Â Uvaro, BWOSS And Garage Capital Announce First Two Recipients Of Impact Scholarship Fund
