Industry IoT Consortium Publishes Guidance On Cloud vs. Edge For AI-based Industrial Applications

The Industry IoT Consortium, part of the Object Management Group, gathers interested technical people together to study and publish guidance on a variety of industrial connectivity applications. This one provides guidance for when to use cloud or edge computing for AI-based applications such as machine vision.

The Industry IoT Consortium (IIC) published the Optimal Use of Cloud and Edge in Industrial Machine-Vision Applications whitepaper. The paper is a brief guide for developers who want to integrate machine vision with industrial cloud and edge computing applications.

“With advances in technology, industrial machine vision is becoming increasingly sophisticated. For example, today you can perform AI vision analysis directly on a camera. You can also process this data on a nearby computer, an on-premises server, or remote data center,” said Daniel Young, IIC Technology Working Group Co-Chair and Senior Manager at Toshiba. “Understanding where image processing should occur is an engineering decision based on many different factors.  For example, cloud computing offers industrial applications flexibility and scalability for machine learning models, while edge computing is best for real-time industrial tasks.”

The whitepaper covers the following sections:

  • Applications of Industrial Machine Vision gives a broad description of how companies use industrial machine vision.
  • Edge Computing in Industrial Machine Vision provides an overview of applying edge computing to industrial machine vision applications.
  • Cloud Computing in Industrial Machine Vision discusses cloud computing in industrial machine vision.
  • Deciding Where the Edge Lies in Industrial Machine Vision talks about typical edge computing configurations that incorporate industrial machine vision.

Metaverse Solutions For Enterprises

Press releases and extensive news coverage provoked some thinking about the Metaverse and its assorted technologies—Augmented Reality (AR), Virtual Reality (VR), eXtended Reality (XR), and Mixed Reality (MR). It’s enough to distort in one’s mind just what is reality. Some psychologists and philosophers think there is no reality outside of what’s in your head. At this rate, they may be right. I even devoted a podcast to thoughts about this.

Mostly I’ve been exploring AR usually in the form of glasses that project the digital over the physical or VR usually in the form of an eye covering totally immersing you in the digital world. I’ve controlled machines while wearing glasses such as Microsoft’s HoloLens and seen training demos in VR. But VR can be on a flat panel, too. My wife the other day was holding her iPhone up and pointing at the walls of her “reading room.” She was visualizing a piece of furniture.

But I am here today to talk about constraints and overcoming them. 

Dijam Panigrahi, COO and co-founder of GridRaster, talked with me the other day. 

We started with constraints. Even AR requires a lot of compute power. And memory. And networking/communication bandwidth. Not to mention an electric power source. It’s hard to fit all that into an acceptable form factor. Rumors were that Apple was about to release its long awaited AR and VR products. The rumors pointed to the need for the wearer to have a battery pack clipped on their belt with a cable to the device. (For those who don’t wear pants with belts, well, I have no idea what you would have had to do.)

Panigrahi told me they started from a different point. They saw the power of the cloud plus the power of new communications networks such as 5G. Add to this advances in 3D CAD. Why, they asked, should designers try to put everything into the wearable device. Why not host the data in the cloud and use advanced networking to communicate with the device.

GridRaster does not design and sell end user devices. It works with any device and cloud service. It has what they call a unified and shared software infrastructure that enables enterprise customers to run AR, VR, XR, and MR applications.

Here are some underlying ideas and technologies:

  • Ultra-low latency high fidelity remote rendering using distributed GPUs for graphics heavy computing for high-fidelity rendering without time-consuming polygon reduction, and wirelessly streaming the solution to headsets, mobile phones, and tablets. 
  • Millimeter precision 3D AI based spatial mapping achieving accurate 3D spatial mapping with high fidelity 3D scene reconstruction, scene segmentation, and 3D object recognition using 3D computer vision and deep learning-based AI running on discrete GPUs on the server. 
  • Auto scaling and deployment using DevSecOps applying gaming tools and concepts in a cloud native environment that allows for agile, secure, and rapid development, deployment, and operations on the cloud/on-premises. GridRaster uses Kubernetes for deployment and scaling. It follows a CI/CD pipeline for deployment of its services into the cluster following the best practices and CNCF graduated projects. This enables loosely coupled systems that are resilient, manageable, and observable, and future proof.
  • Easy API-based integration open architecture approach enables a frictionless onboarding and seamless integration with existing content formats and provides future proof cross-platform support. This also allows the platform to integrate with other systems to share data and allow for interoperability.

And some use cases:

DESIGN & ENGINEERING

Enables Real-time Collaboration for Rapid Prototyping

  • Enables quick iteration on the ideas and concepts 
  • Clearer communication among team members
  • Quick decisions. 
  • Precise overlays of virtual models on real-world on any commercially available mobile devices, HMDs, smartglasses and PCs in real-time.

REMOTE MAINTENANCE, REPAIR AND TRAINING

  • Enables photo-realistic visualization and remote collaboration.
  • Create a virtual environment close to real world settings, along with photo-realistic product visualization and real-time collaboration, so that the most effective environment can be created for remote maintenance, repair and training.

LARGE SCALE SIMULATIONS AND TRAINING

  • Manages ultra-realistic mixed reality simulations.
  • Combines the best of the gaming and traditional simulations to provide a massive multi-user and multi-platform ultra-realistic large world simulations in AR,VR and MR. 
  • Provides a cloud-based agile and secure deployment and operation that distributes complex computations across compute server nodes and handles scaling in real time using Kubernetes.

AVEVA World Announcements

I had committed to a couple of events when the invitation to attend the AVEVA conference in San Francisco in November came my way. Many of my colleagues went out there thinking they had better opportunities to sell advertising than the alternative. Subsequently I received two announcements from the event. Both relate to data.

  • Constellation Energy chooses AVEVA solution to enable easier, faster, cheaper data analytics
  • Vision for connected industrial ecosystems revealed at AVEVA World in San Francisco

Weird thing to me is how marketing over the past few years regards data as something new. Way back in 1976 while working for a manufacturing company, I was moved from a position in operations to one in product development with the principle role as sort of “data czar.” I learned as far back as then about how crucial it was to have verified data in a form usable by all areas of the corporation. I was sort of at the nerve center for a few years.

We just have technologies for compiling, storing, verifying, and visualizing data that I couldn’t even dream about back then.

Constellation Energy for data analytics

AVEVA announced a partnership with Constellation Energy, provider of carbon-free energy, to give third-party analytic vendors secure access to select real-time data from Constellation’s operations.

Constellation will implement the latest release of AVEVA Data Hub, a cloud-based data management SaaS solution, which will help the company accelerate collaboration with trusted analytic partners and implement learnings more quickly, increasing the return on its analysis projects.

Currently, industrial operators, who are focused on data-driven solutions and decision-making, use a variety of methods to make real-time data available to internal and external teams. These methods are time-consuming to develop, maintain, coordinate, and some of them introduce security vulnerabilities that increase business risk. With IT staff in short supply, companies find themselves unable to respond quickly to new data requests or test out new solutions.

Constellation (formerly Exelon Energy), a long-time user of AVEVA PI System, decided to expand their data management solution into the cloud to easily share wind turbine data with a third-party analytics vendor. The company uses AVEVA PI System to collect, enrich, store, and manage sensor and time-series operations data in real-time. It then transfers that data natively to AVEVA Data Hub, a SaaS solution, where they can set up and manage select data sharing with authorized users, applications, and analytic tools both inside and outside the company. The new cloud-based addition to their data management solution is ready to use immediately; users can begin analyzing and processing data to achieve deeper insight in hours, rather than days or weeks.

Vision for connected industrial ecosystems

AVEVA showcases how industrial organizations are using real-time data to connect teams, empower them with data-led insights that speed up decision-making and unlock business value.

“We are witnessing the birth of an industrial universe that is completely connected, enabling a new kind of collaboration across colleagues, suppliers, partners, and customers,” said Peter Herweck, CEO at AVEVA.“Taking a data-centric approach empowers teams by connecting different players across the entire industrial ecosystem. This in turn transforms value chains into agile, profitable, sustainable networks. It is what we at AVEVA mean by the new, connected industrial economy.”

A recent survey, commissioned by AVEVA, of 650 senior international business executives across the chemicals, manufacturing, and power industries in North America, Europe, and the Middle East, found that 87% said they plan to increase their organization’s investment in industrial digital solutions over the next 12months.

Herweck added: “When you bring your data together and apply analytics so that you can visualize it in context, you unlock new ways of working. We are seeing leading companies like Shell and Worley breaking down data silos, building digital twins to deepen collaboration, drive transparency, and deliver actionable insights that enable their teams to work in a smarter and more connected way.”

AVEVA World has shown how leading companies such as Kellogg, Barry Callebaut, Pfizer, Dominion Energy, and Henn, starting to put in place the building blocks of these connected industrial ecosystems.  As the adoption of cloud-based industrial software becomes more widespread, organizations will be able to engage experts within and beyond their enterprise to deliver on innovative capital projects, optimize the operations lifecycle, accelerate decision-making, and reach sustainability targets that drive responsible use of the world’s resources.

Automation Fair Attracts Large Crowd, Exhibits New Technology

There was Automation Fair and then meetings and then Thanksgiving. Plenty of time to digest everything I learned and experienced at Rockwell Automation’s event. They skipped 2020, of course. The 2021 event in Houston was abbreviated. This one held at Chicago’s McCormick Place resembled all the pre-Covid events. More than 15,000 attendees, lots of presentations and learning opportunities, and many Rockwell and partner exhibits.

I wrote a quick update with the themes I picked up. Here is a recap.

Themes

  • Emphasis on partners working together—specifically Festo, Zededa, Stratus
  • Connected Enterprise is still central, connectivity is everywhere
  • Cloud is everywhere
  • Use technology the right way
  • Enable humans to make better decisions

The big three things Rockwell spokespeople referenced were FactoryTalk Design Hub, FactoryTalk Edge, and cloud. The cloud discussion involved Design Hub, Plex, and Fiix. Plex, the cloud-based MES software replaced PTC as the focus of software discussion. PTC’s stand, though, with ThingWorx and Vuforia, was constantly busy—so much so that I never squeezed in for a talk. Fiix is the cloud-based CMMS package. I think these were meaningful acquisitions that will add value now and in the future.

Specifics — Rockwell Automation Products

FactoryTalk Design Hub

Transform automation design capabilities with a more simplified, productive way to work powered by the cloud. There are five core applications.

  • FactoryTalk Design Studio—cloud-native software product built from the ground up to improve system design efficiency. Teams can collaborate with automated tools to share and merge changes, and project sizes can scale dynamically with support for multiple controllers in a single project.
  • FactoryTalk Optix—addition to the visualization portfolio, the first cloud-enabled HMI product to be launched within FactoryTalk Design Hub. 
  • FactoryTalk Twin Studio—end-to-end automation design solution where users can design, program, simulate, emulate, and virtually commission in one cloud environment.
  • FactoryTalk Vault—secure, cloud-native centralized storage for manufacturing design teams. 
  • FactoryTalk Remote Access—enables secure connections to equipment. 

FactoryTalk Edge

An intelligent edge management and orchestration SaaS platform with an edge application ecosystem – based on zero trust security and open industry standards – accelerating digital transformation for industrial customers.

FactoryTalk Edge Gateway (FTEG)

Connects operational technology to informational technology. FTEG tools scan the EtherNet/IP network to discover devices. 

FactoryTalk Smart Object (FTSO)

A simplified way to organize data for easy collection by the controller and subsequent transfer to IIoT systems. The FTSO Config Tool creates the base code for the tags the user selects to be included in data models.

On Machine Products

  • ArmorKinetix distributed servo drives provide the Kinetix 5700 platform in a compact, On-Machine form factor.
  • ArmorBlock 5000 I/O—Distributed I/O blocks with IO-Link technology
  • Armor PowerFlex—A new generation of On-Machine VFD motor control solutions that provide an integrated, near-motor solution where reducing installation time and cost are most critical.

Specifics — Partners

Festo

  • With the Festo CPX-FB36 bus node now recognized by FTEG, basic diagnostic information from smart pneumatic devices such as Festo valve terminals and energy saving pneumatic devices are easily accessible.
  • Add-on instructions can be used by FTSO that access timestamp values for the extension and retraction of pneumatic cylinders. This information is accessible to FTEG.
  • Data collected by FTEG can be shared with Festo AX software, the new AI solution from Festo that can improve machine utilization and quality, lower waste, and ensure energy optimization. Festo AX provides a bridge between advanced analytics arising from operational technology, such as that collected by FTEG, and IT-based business intelligence. Festo AX can run directly on the system (on-edge), on servers (on-premises), or in the cloud.

Zededa

Zededa announced a supply agreement with Rockwell Automation to provide distributed edge management and orchestration capabilities—incorporated as part of FactoryTalk edge management offering.

Ignition Adds Cloud Edition Plus Many Updates

I think that hybrid might just be the future as IT managers review the costs and constraints of depending too much on cloud vendors (AWS, Microsoft, Google). Customers of Inductive Automation’s Ignition platform must be asking for a public cloud implementation, though. 

I’ve highlighted a blog post by Dante Augello of Inductive discussing some of the reasoning and features of a Cloud Edition coming in 2023. Following that announcement are notes about the latest update to Ignition—8.1.22.

Ignition Cloud Edition

  • Many hybrid architectures have taken advantage of the cloud over the years, allowing controls applications to run on-premise while communicating and sending data to the cloud in order to take advantage of virtually unlimited storage and computing power. 
  • This connection to the cloud also offers better integration with artificial intelligence, machine learning, and other cloud-native technologies.
  • The goal of Ignition Cloud Edition is to enable a hybrid architecture that connects one or more on-premise gateways and numerous edge gateways that send information to a Cloud Edition gateway for enterprise-wide data aggregation and monitoring.
  • This new product will differ from the standard version of Ignition in three key ways: distribution, purchasing, and features.
  • Unlike standard Ignition, Ignition Cloud Edition won’t be a service provided directly through Inductive Automation, so it won’t be downloaded from the Inductive Automation website. Like other cloud-based applications, it will be made available through your preferred cloud infrastructure marketplace.
  • Since Ignition Cloud Edition will be purchased through a cloud infrastructure marketplace, there is no need to buy an upfront license as with standard Ignition. This allows for elastic deployments, scalable payment options like pay-as-you-go, as well as the ability to adapt compute size and number of instances as needed.
  • Ignition Cloud Edition will run in the cloud, so it won’t be ideal for direct data acquisition from plant-floor PLCs. For this reason, it won’t have Ignition’s original device drivers built in. Instead, Cloud Edition will have a collection of cloud connector modules for cloud-native technologies such as document databases, message queues, and key-value stores like MongoDB, Kafka, and Redis.
  • A key benefit of Ignition Cloud Edition’s infrastructure is elasticity. If you find that you need more or less computing power than you anticipated, you will be able to scale the infrastructure much more easily than with an on-premise infrastructure.

Ignition 8.1.22: New Configuration Explorer, Enhanced SVG Importer, and CSS Stylesheet Resources

This blog post by Aaron Block gives points about the latest update.

  • Ignition 8.1.22 arrives with major upgrades to Perspective, the Gateway Network, and redundancy, plus improvements focused on general quality of life.
  • Perspective gets the majority of attention in Ignition 8.1.22, with three big updates: a new Configuration Explorer, SVG Importer enhancements, and resources for CSS stylesheets.
  • The Configuration Explorer locates active bindings on any particular Perspective view. Now in Ignition 8.1.22, simply right-click and choose “Configuration Explorer” to display the location, type, and state of all enabled bindings. Embedded views and root containers are also represented in this overview. 
  • In prior versions of Ignition, it was sometimes difficult to import complicated SVGs with intricate shading or other non-supported components within the raw SVG file. 8.1.22 offers more support for these complex SVGs, as well as most common SVG elements.
  • The “convert to drawing” function stays true to its name, allowing users to simply choose the function in the right-click menu and convert existing components into SVGs.
  • Rounding out the trio of major Perspective updates is an advanced project-scoped CSS stylesheet resource and resource editor. Users can now conveniently add their own CSS stylesheets into the Perspective project itself instead of spending time inserting CSS through the gateway’s style directory.
  • And there are many more updates. Check out the blog for more detail.

Rockwell Automation Announces Edge-to-Cloud and Upgraded Asset Management

Here are two interesting announcements from Rockwell Automation this month. DataOps and asset management subscription service—both technologies for the future.

Edge-To-Cloud

IT companies have been touting edge-to-cloud architectures for several years. I’ve followed several of those companies until they decided there’s not enough revenue from the industrial market to satisfy them. Industrial technology suppliers have been notoriously slow to adopt new technologies. But the pace is picking up. 

In my next post after this one, I’ll discuss another industrial company who has discovered edge-to-cloud as a promising architecture of the future. Rockwell Automation beat them by few days. I have two announcements from Rockwell from last week revealing that company’s continued advance into software and IT-related technologies. Some executive sources in the industry have told me that Wall Street is driving much of this strategy by valuing software companies over hardware companies. 

Check out Rockwell’s investment in and subsequent partnership with PTC that enhanced its software connectivity of automation to IT. Then acquisitions of Plex and Fiix to get into the cloud. Now a partnership with Cognite for a data hub—yet another key component.

  • Strategic partnership to develop a unified, edge-to-cloud industrial data hub offering
  • Combines Rockwell’s FactoryTalk software with Cognite’s Industrial DataOps platform Data Fusion [Note: I’m on record somewhere predicting DataOps as a cool tool for the future.]

My Equipment Subscription

I always forget that often when Rockwell talks asset management, it’s not the same thing as when a process automation company discusses it. In Rockwell speak, this means keeping track of Rockwell assets in a manufacturing company.

  • My Equipment Digital enables digital collection and updating of industrial automation asset data through the use of network-based devices. 
  • The digital collection method includes firmware information and a one-time vulnerability assessment.
  • Customers can self-manage and track their installed base automation assets
  • The self-manage feature can be added to an Installed Base Evaluation (IBE service) and is included in the My Equipment Digital and My Equipment Managed offerings.
  • Mitigate network risks if legacy products and/or unmanaged switches are found
  • Review modernization potential to determine risks and productivity improvement opportunities by identifying which older products may be less expensive to repair than to buy new
  • Provide adequate remote support coverage by confirming that the most prevalent technical segments are covered
  • Reduce costs through the optimization of asset and spares strategies by comparing storeroom data to the installed base, and the identification of warranty savings

Follow this blog

Get a weekly email of all new posts.