The More We Get TogetherJuly 16th, 2021 by Nathan Hobbs
With increased connectivity, software sings its way to singularity
By Drew Vass
Artificial intelligence (AI); Industry 4.0; the Internet of Things (IoT); the Industrial Internet of Things (IIoT); there’s no singular definition for any of these concepts, says A.J. Piscitelli, project manager for FeneTech. “That’s one of the reasons why we came out and said, ‘Okay, we’re just going to say digital factory and it’s going to be an all-encompassing thing’—the whole interconnected workshop.”
No matter what you call it, from a developer’s perspective it all boils down to one thing: the more connections and data they ascertain, the more intelligent their software is capable of becoming. For this reason, in recent months fenestration-related software companies have focused on extending the reach of their data gathering to as many points as possible throughout industry processes.
“Our primary focus is building applications and services that can take the data that’s coming from all these connections and actually do something with it,” Piscitelli says. From various software modules to machines, “we want those to be able to submit their data, so we can render it—do data analysis and aggregation, then present useful information back to our customers,” he says.
As a result, over the past couple of years, FeneTech has worked to deploy FenML, a baseline communication standard that leans on IoT (or whatever you choose to call it) to allow the machinery and software produced by various manufacturers to “speak” the same language. Other developers have worked on everything from mobile apps designed to scan bar codes and near field communication (NFC) labels, to artificially intelligent, cloud-based applications that produce augmented reality experiences or business insights. So far, humans hold the final call in software-driven, artificially intelligent processes, but there will come a day, developers suggest, when machines will keep that information to themselves—if humans can learn to fully trust applications. That’s the goal that software providers are slowly inching toward.
Brains of the Operation
For developers of enterprise resource planning (ERP) software, advancing to AI only makes sense, some experts suggest. Companies such as A+W Software, FeneTech, Paradigm and Woodware already offer programs that gather, process and provide data and insights across all areas of business and manufacturing. And because most ERP platforms are now cloud-based, or at least cloud-connected, “You can be sitting on a beach,” while the software monitors certain processes and you post final decisions, says Brad Carter, sales manager for Woodware Systems.
The first tier of AI includes allowing cloud-based programs to automatically analyze enterprise-wide data in order to trigger automatic reactions. Through functions commonly referred to as “if-this-then-that,” or IFTT, software such as Woodware’s Workflow application can automatically monitor events in the software’s database, then notify the proper users by email with suggestions. The system also automatically triggers events based on certain detections and user-defined schedules. And there are plenty of compelling reasons to move forward with these processes. Though computers may not be more powerful than the human brain—yet—they’re infinitely more capable of crunching big data, developers say, and therefore more than capable of generating accurate, logic-based decisions.
For example, FeneTech offers a capacity planning module for its ERP platform that examines all current orders, production demands and capacity, even measuring the capacity of individual work centers in order to automatically determine when and where actions should be placed within a company’s schedule. With those data points, a person can make the same decision; with larger sums of information, however, machines can do it faster.
These days, most ERP modules “present all of the information to the user to say, ‘This is what we think you should do,’” Piscitelli says of FeneTech’s software. “But ultimately you have the final say … in terms of the user, they’re basically just saying, ‘Okay, yeah, go ahead and do it.’”
Those are the types of human-machine interactions that are necessary to build trust, says Kari Tamminga, product owner and leader of a data science team for Paradigm. And it will take many more of those experiences to nudge the industry toward full AI. First, companies must see that artificially intelligent software can and will work as expected. “We’ve done our research and we know that’s what people care about,” Tamminga says. Otherwise, the attitude is, “If it doesn’t work, I’m not using it,” she adds.
In recent months, Tamminga and other developers at Paradigm have worked with independent computer scientists to develop software that’s capable of doing something that comes fairly easy to humans: identifying fenestration in building envelopes. To get software to recognize windows that are partially blocked by tree branches, or doors that are darkly shaded by overhangs requires deep learning for software, says Matt Davis, product owner for Paradigm.
“Yesterday, we sat in a room for two hours with one of the smartest people I’ve ever met—a gentleman who’s from the Department of Bio Statistics and Computer Sciences at University of Wisconsin,” Davis says. “He is a world-renowned data scientist.”
By linking a cloud-based system to “millions of photographs” of houses and other buildings, Tamminga says her company has been able to train software to accurately analyze pictures captured and uploaded by a smartphone app to identify doors and windows. By linking that information to real-time product specifications stored by Paradigm’s Omni and Nexus software platforms, the program suggests and accurately virtualizes replacements. Through cloud-based computing, the entire process, company officials say, takes just seconds. The goal includes making the system available to manufacturers to be integrated into their websites.
“In the broadest view, it’s about looking for patterns in large amounts of data,” says Greg Holmes, a developer for Soft Tech about AI. In order to do that, it isn’t so much about designing every system from the ground up, says Davis. Instead, developers look to leverage existing software-based intelligence and algorithms.
“When you’re talking about a starting point, it’s exactly that,” he says. “Let’s find the smartest people in the world who are dealing with issues like this. Let’s leverage models that they’ve created, or thinking they can provide us to get us there.”
Once systems are fully trained to connect with manufacturing and business processes, “we can provide [AI] with full-size data sets and ask questions of it,” Holmes says (in much the way consumers ask questions of Amazon’s Alexa, only less “if-then-oriented” and programmatic). For manufacturing, those advancements are about gaining quality and efficiency, he says.
“The problems that AI will be able to solve include answering key questions around how to best optimize workflows, to maximize production and when to schedule maintenance, so as to get the most out of the lifetime of equipment,” Holmes explains. And there are some compelling reasons for manufacturers to allow software providers to compare data across company lines, some developers suggest. For instance, by recognizing patterns for technical failures among certain equipment, relative to such factors as material types, demand and number of production cycles, artificially intelligent software might be able to establish and suggest when one manufacturer should expect to see downtime, based on data collected from throughout the industry.
So far as how such comparisons might happen, developers would need to, “take all that data that’s already staged and ready to go and say, ‘Alright, let’s start doing some conditional monitoring and some machine learning,’” says Piscitelli. “We can take a certain model of machine and say, ‘Well, there’s six or seven different customers that are running that machine and we want to find out: When are these machines failing? When are they throwing up errors? What are the conditions that superseded that?’” he suggests.
While all of the developers and company representatives interviewed for this article reiterate the extents they go to in order to protect their customers’ data—and keep it strictly separate—in the years ahead, those comparisons could happen with the permission of manufacturers. The concept of data comparison could also extend beyond the industry, suggests Chris Kammer, marketing coordinator for A+W, by tying ERP platforms to third-party data. For example, he says, “the smart city and the smart building communicate.” By connecting their data to ERP systems, AI could suggest, “Okay, the most optimal delivery route is this today, on a Tuesday, and not what it was last Tuesday, by the way,” he says.
Bigger is Better
With AI, the larger the data, the more intelligent the insights that can be provided, Holmes explains. “These complex problems involve a huge number of variables and factors, which is where the power of machine learning starts to come alive,” he says.
Before any machine learning can happen throughout the fenestration industries, first machines must be enabled to communicate. For this reason, developers have worked to create more data points and integrations.
FeneTech’s FenML (short for Fenestration Manufacturing Language) is a cloud-based, baseline communication standard that allows equipment to capture and relay data, such as alarms, temperatures, power consumption, vibration, operating modes and cycle times to other connected systems, where they can be analyzed for production and other business decisions. Meanwhile, to help digitize items not equipped with software and communication chips, other developers are working on smartphone-based apps designed to read and write cloud-based data via product labeling.
A+W and Veka have spent recent months developing apps designed to integrate with ERP software, including Smart Companion, an app by A+W that utilizes the built-in camera on iOS- or Android-based devices to scan barcodes. Unlike standard, handheld scanners, smartphones are capable of storing and retrieving meaningful amounts of data directly to and from the cloud via internet connections, explains Kammer. As a result, “You can keep track of everything that’s going on with every piece that heads out the door or that’s going through the factory, right where it is,” he says. Currently, the app is available only for the glass industry. GED introduced a similar system in 2018 that utilizes RFID technology—automatically detecting and tracking window assemblies.
Meanwhile an upcoming technology developed by Veka will take the concept of cloud-based data and product labeling beyond just manufacturing.
In July, Veka announced a new venture called Digital Product Solutions Corp. (DPS), for the purposes of developing and providing fenestration-related services and technologies, says Steve Dillon, Veka’s marketing director. Among the first projects to be developed is WIN (short for window intelligence). Currently in beta, the patent-pending system utilizes near field communication (NFC) tags placed on door and window products to retrieve and store cloud-based information via the NFC reader on an average smartphone. By linking those NFC tags to data stored by manufacturers’ ERP systems, Dillon says the company will provide information over a suite of mobile apps—including such things as testing data, service reporting or warranty claims.
By extending real-time data collection throughout manufacturing and business processes, software developers say their systems will slowly but surely become capable of collecting enough information to make immediate decisions. Like a central nervous system, machines and mobile devices will provide feedback for artificially intelligent reactions. So far as when workers and companies will choose to sit back and trust those processes, “I definitely think we’ll get there,” Piscatelli says. “It’s just like working with any person. I’m going to let you make some decisions, but I’m going to verify that you’re making good decisions, before ultimately giving the go ahead.”
WoodWare includes a fully integrated set of applications designed for managing door, window and millwork operations. Each of the software’s features handles one aspect of processes, but cross communication facilitates information flow for ordering and inventory tracking through business analytics. The company’s WorkflowXT and Business IntelligenceXT applications work together to provide intelligent insights into things like cost control, customer service and business optimization. Workflow monitors your data and initiates action based on key performance indicators.
In May 2019, Andersen Corp. announced an augmented reality-based tool for its Renewal brand of replacement doors and windows that utilizes an iOS-based app for the iPad to overlay replacement products over existing doors and windows. The tool, officials said, is “different from other augmented reality tools of the past that plug in an animated, cartoonish picture of a product in a room,” instead placing images of real-life products. The software utilizes vertical plane recognition technology.
Developers at Paradigm have worked to develop a web-based platform that utilizes artificial intelligence for identifying fenestration. Announced in April and currently in Beta testing, View utilizes millions of images to learn how to correctly identify doors and windows in up-loaded images. The application then draws on product specifications stored by manufacturers in the company’s enterprise resource planning software to produce real-time images of what various products would look like when installed.
The system is expected to go live in fall 2019.
Introduced in September 2018, FeneTech’s FenML (short for Fenestration Manufacturing Language) is a cloud-based communication platform that’s designed specifically for the fenestration industry. Through a “plug and play” format, various machines, work-cells and software applications communicate information like cycle times and maintenance notifications. Those integrations also allow for the use of historical data across ERP platforms.
In July, Veka Corp. announced the formation of Digital Product Solutions Corp. (DPS). Among the first products scheduled for roll out is WIN (short for window intelligence), a product identification system that’s based on near field communication (NFC).
Currently in Beta testing, the patent-pending system creates a “digital double” of products based on data gathered from materials and production. A web-based program is tied into fabricator’s enterprise resource planning (ERP) software to gather and build a complete dossier of doors and windows. The company is aiming for full commercialization by the end offirst quarter or beginning of second quarter 2020.
Smart Companion, a new mobile app from A+W utilizes iOS- and Android-based smartphones to scan barcodes. Designed to replace handheld scanners, the app connects to enterprise resource planning and production (ERPP) software in order to track and digitize product-related information. The app can scan numerous barcodes at once or in sequence (e.g. edge labels), providing real-time information, such as the location of various components within a facility, or the contents of racks. If an employee isn’t connected to the network, data is recorded securely in offline mode and transferred as soon as there is a connection. The app is currently only available for the glass industry.
Drew Vass is editor of [DWM] magazine.
To view the laid-in version of this article in our digital edition, CLICK HERE.