- Companies are facing challenges associated with an AI system that doesn’t have enough, or the correct, data to provide benefits.
- When the system works well, management can then gain insights on the way production work or material flow is taking place.
Factories are introducing AI into their product design, assembly automation and robotics, as well as exploring ways to make general efficiency improvements. And some are leveraging the rich data provided by RFID solutions to enable that effort.
In fact, the percentage of those using RFID will rise to 66 percent within the next five years based on the plans of manufacturing leadership surveyed this year in a Zebra Technologies study.
Zebra has seen manufacturer customers benefit with real-time locating system (RTLS) IoT data and AI such as Yamaha G3 Boats which has maximized its efficiency through visibility of the manufacturing process. An agricultural company in California has leveraged Zebra technology-based data to direct robotic watering and nutrient distribution to plants, thereby cutting water use.
But whether using IoT technologies (like RFID) to leverage AI is a good strategy requires a close look at the quality of that data. Too often, companies are facing challenges associated with an AI system that doesn’t have enough, or the correct, data to provide benefits.
Connected Factory Growth
One of the findings of the Zebra study— known as The Rise of the Connected Factory— is that collecting the right data is essential to manufacturing leadership. Enrique Herrera, Zebra’s industry principal for manufacturing, pointed to some of the study’s key results in a recent conversation about the role RFID and RTLS data are playing for manufacturers looking at AI solutions.
Zebra Technologies commissioned the manufacturing survey, conducted by Azure Knowledge Corporation. It included 1,200 online surveys among C-Suite executives and IT and OT leaders within various types of manufacturing sectors including automotive, electronics, food and beverage, and pharma and medical devices. Respondents were surveyed in Asia, Europe, Latin America, and North America. The results offer a glimpse into the IoT adoption planning and challenges for those with manufacturing sites.
Among its findings, the study determined 31 percent of manufacturers are already using RFID to increase efficiency and production quality. That adoption is expected to rise to 66 percent by 2029.
Gaining a View into Movement
One of the themes of the vision study was the set of pillars that make a connected factory possible: including visibility into operations, visibility into the equipment, and understanding of the movement of materials, products and people on the production floor. That visibility of movement can help identify nonconformance that could affect quality of finished products as well.
One example can be simply capturing the movement, of goods and people, that can feed into an AI system to better understand waste. If staff members are rushing around the factory floor, looking for a tool, or if a product moves from one station to another and then reverses course, an RFID or RTLS system can collect those actions into a set of data that AI is designed to analyze.
When the system works well, management can then gain insights and recommendations to better calibrate or recondition the way that production work or material flow is taking place.
Filling in the Vision Gaps
Manufacturers still struggle with a key challenge however: the lack of visibility into their operations. Despite a wide variety of visibility initiatives and technologies, the companies that produce the world’s goods still struggle to see what is taking place in some meaningful ways.
“It’s the granularity into certain aspects of the business,” said Herrera, that still challenge business managers in terms of actions taking place onsite as well as what occurs outside the four walls in the supply chain as well.
One example, Herrera pointed out, was a finding that didn’t surprise him: visibility in the warehouse and the flow of assets and materials through those warehouses is becoming more important for understanding what’s happening on the factory floor.
Examining the Data
So as manufacturers look to future automation and technology improvements, the AI adoption conversation focuses on granular visibility. Without it, “AI will struggle with some of that underlying data,” if that data is not properly strategized, Herrera said. Not only can data lack the proper granularity, but it can also come from a variety of sources that are not connected.
AI and its Role in RFID and IoT – the fifth in an ongoing series exploring how AI is impacting the RFID and IoT industries.
AI Drives Efficiency for Automotive Management
For instance, a RTLS system may be tracking conditions related to equipment in use at a production site, while RFID is tracking the flow of materials or work in progress. “I think the data-centric and data engineering piece is fundamental,” he said.
That means companies often benefit most from building their data stream in such a way that it enables AI to work properly.
“That’s a sticking point for many,” Herrera said.
Using AI in Conjunction with Frontline Worker
The Zebra survey found many manufacturers are approaching technology adoption in general as a collaborative tool that helps automate but that also can work with frontline workers. The conversation, Herrera said, “is not all 100 percent ‘let’s automate and remove the frontline worker’,” but instead an approach that puts robotics and AI based technology into the hands of frontline workers.
“Manufacturers are looking for a way to not eliminate frontline workers; they are hoping to augment what they do with technology,” he said.
Herrera pointed out that AI is a broad topic with numerous subsets, including machine vision, natural language speech, robotics—all of which can fall under the AI umbrella.
Many initiatives today center around machine vision and robotics to improve some of the existing processes. But these systems rely on good data. With that in mind, Herrera noted “RFID technology is very relevant to give a voice to the mobile asset or person or tool.”
Where to Start
One challenge revealed by the survey was where to start. It’s a question, Herrera said, of helping manufacturers expand beyond pilot purgatory—that place of testing without enabling an actual deployment.
Companies need to make sure their business metrics, objectives and imperatives are aligned, and then begin prioritizing the data to be gathered, based on those business outcomes. In many cases, the approach means getting the proper data collection in place and then introducing AI.
Additionally, to ensure an IoT and AI system that leverages automation along with skilled labor, companies need to train those in contact with it in order to take full advantage of the information being collected and analyzed.
“That means making sure that the data is engineered in a way that enables AI,” Herrera pointed out, “and that outcome is realized not just by the person creating the models but making sure that that model actually goes back and is implemented in an operating environment to drive business value.”
If a company starts with the AI technology, with a plan to next generate the data to try to build the necessary model, frustration can result. One example is employing AI to benefit one part of a factory or one worksite. If a company then wanted to roll the AI system across hundreds of factories—or numerous parts within one factory—they may not have the infrastructure and the architecture in place to enable that expansion.
Early-Stage Deployments
Currently, most AI and RFID or RTLS deployments are still in the earliest stages for manufacturers. “People are using RFID or RTLS in a variety of ways even back to the warehouse to improve operations. As far as introducing AI into that, that’s still early going,” said Herrera.
He added that ultimately “getting a handle on their data is pretty important,” and some early pairing with AI is helping companies in tasks such as reduction of inventory or optimization of movement.
Most early adoption is centered around management of inventory and improving the flow of supplies and finished products, or for predictive maintenance. If an AI system can be used to analyze disparate data from multiple systems, factories will gain insights that weren’t possible before. That can include identifying the relationship between operational flow and efficiency and the uptime of the machine, for instance.
However, there’s a wide variety of maturity among Zebra’s customers, Herrera said. Some may have bleeding edge technology already in place, while others are still considering how they will begin using the data that’s being collected from a legacy system.
“There’s still a vast majority of manufacturers that have a long way to go just in the data collection —some feel that they they’ve collected a lot of data and they’ve amassed it in a data repository and they think ‘let’s take AI onto it and see what we can find’,” said Herrera
For some, however, he added, “I think that’s where they’re realizing it’s not as easy as they thought.”