USG Corporation is the largest manufacturer of gypsum products in North America. With a history of adopting new technology to enhance efficiency and quality across its network of over 50 manufacturing sites, USG implemented an initiative to optimize physical processes with the latest AI technology.
This initiative combined USG’s experience in manufacturing with machine learning, specifically vision AI, to enhance quality checks and minimize downtime. As a result of this transformation, USG has experienced measurable improvements in quality, yield, and cost efficiency through smart factory automation and readily accessible insights for facility staff.
At the heart of USG's strategy is a philosophy that prioritizes using new technology to connect people, improve their access to data, and reduce their time spent on repetitive tasks.
“I call it the hands-off, eyes-focused, minds-empowered approach to manufacturing,” said Lou Stocco, Director of Manufacturing Advanced Analytics at USG. “By automating routine tasks with AI, like visual quality inspections, staff members can spend less time with their hands or eyes focused on dull and sometimes dangerous work. Instead, they can focus their minds on pursuing higher value initiatives that ultimately improve success for our customers.”
When implementing the above philosophy, vision AI has been a cornerstone of USG's strategy, addressing key processes like quality control and unplanned downtime.
For example, in the drywall production process, operators at USG needed to identify products that were misaligned as they moved down the manufacturing line. Previously, when a misaligned product went unnoticed, it would snag on machinery causing a jam or pileup. These jams can create hours of downtime, resulting in reduced throughput, time spent clearing the jam, and additional costs from reworking damaged items.
To detect misaligned products before a jam occurred, the team at USG used Roboflow to develop a purpose-built vision AI system. The system continuously monitors drywall boards on the production line, evaluating the dimensions and angle of each one, and identifies when a board might cause a pileup. The AI system is integrated with industrial machinery, so it can automatically reroute products or pause the production line when an issue is detected and notify staff.
“Because of the reliability and performance requirements in our manufacturing environment, we wanted to develop an in-house AI model capable of running on hardware in our facilities,” said Nick Urena, Computer Vision Engineer at USG. “With Roboflow, we were able to create a highly accurate model, based on data we own and tailored to our production requirements. Because it runs on local hardware, it’s capable of analyzing images in a fast-moving manufacturing line, even if our internet provider suffers an outage.”
USG also deployed a custom AI system to automatically evaluate the quality of products and improve the consistency across their network of facilities. This vision-based solution identifies imperfections and helps teams at different sites maintain consistent standards.
“Creating a standard measurement is incredibly important when checking product quality at dozens of sites,” Stocco explained. “With the AI-powered system we built with Roboflow, we established accurate quality benchmarks. This allowed us to pinpoint the characteristics that define the best drywall product and enabled our team to shift from subjective judgment calls to actively elevating quality consistency.”
With teams spread across 50 manufacturing sites, another priority for Stocco’s Advanced Analytics team at USG was giving teams access to accurate data for decision making and resource allocation.
With real-time production data coming from the quality control and jam detection systems, Stocco’s team built a central reporting dashboard to consolidate information from different facilities. The dashboard allows the analytics team at the corporate headquarters to identify anomalies, such as quality deviations at particular manufacturing sites, and proactively offer assistance to distributed teams.
“The production metrics we get from these systems helps us initiate conversations between distributed teams,” Stocco said. “When we notice a process is deviating from the norm or perhaps a better formula was discovered by a team at another facility, there’s an opportunity to initiate a human-to-human conversation and make sure our teams are working together toward the ultimate goal of delivering the best product to our customers. The vision AI solutions we developed with Roboflow have been a core component of connecting our teams and unlocking their ability to collaborate.”
USG is the leader in manufacturing high-quality building materials designed to drive efficiency and innovation across the construction industry worldwide. We exist to help our global customers more easily, safely, and affordably create sustainable and accessible spaces for all. Our wall, ceiling, joint treatment, flooring, sheathing, and roofing products are used to build many of the world’s most iconic structures as well as airports, hospitals, schools, offices, and the places people call home.
