Artificial intelligence (AI) is no longer the stuff of distant science fiction — it is, in fact, knocking at the door for many manufacturing industries. Food and beverage processors may have cracked the door open to check things out at this point, but they need to be ready for the AI invasion and all the benefits it could bring.
For processors that might categorize themselves as “not yet ready,” there’s still time to prepare. Jorge Izquierdo, vice president of market development for PMMI, The Association for Packaging and Processing Technologies, says the food industry as a whole remains at a starting point when it comes to AI.
“We’re just starting, and there are many different technologies, new and old, that can benefit,” Izquierdo explains. “It’s already here, but more is coming, and we should be open-minded and understand that there are very good opportunities ahead.”
Industry currently stands at the point where AI applications can be cost-prohibitive, and therefore, processors need to do a deep dive on the return on investment (ROI) to find the best opportunities. Highly variable processes and tasks, for one, are prime candidates for initial improvement via AI.
“One good example would be e-commerce, where a processor is trying to pack several different products with different shapes and sizes into the same package for shipping,” Izquierdo says. “In a process that deals with that amount of variability, there’s a good chance AI can help.”
AI can help streamline decisions on type or size of package, how to place them in the package, and how to ship that order — and then it can communicate with the equipment to actually package the items the most effective way possible. For many companies this process remains a manual effort.
A ‘vision’ of smoothing out variations
Early opportunities for AI implementation go beyond packaging of e-commerce shipments. Processors of products that have significant variation from item to item will be the earliest beneficiaries of AI applications, Izquierdo believes.
“If you think about a cookie factory, for example, where all the cookies are the same shape and size with minimal variation, AI will still help, but that’s much further down the road,” he says. “However, with meat and poultry or fruit and vegetables, the product variability in shape, size and even color is very significant, and that’s an area where AI can make the process much better.”
AI technology, for example, can analyze the color of the products and determine the exact mix of gases when packaging them to maintain and manage product consistency when it reaches the consumer. For fruits and vegetables, AI can help processors control just how ripe the produce will be when the consumer receives it.
Vision system capabilities can take a significant step forward with the help of AI. This technology already has transformed processors’ ability to sort, analyze and direct products within plants to provide improved product consistency and quality, as well as food safety in many cases.
Standard vision systems require the processor to “train” the system on product differences. Products with minimal variations make calibration slightly easier, but for products with a virtually infinite set of variables — such as animal protein or produce — calibration of the vision system becomes less of an exact target.
“Rather than trying to give the vision system a full sample of all options and let it run, processors can use AI here to allow the system to look at each product and learn how the operation should handle it,” Izquierdo explains. “Many of the pick-and-place applications are still being done by hand because of the random shape and form of the product, but this goes beyond just telling a pick-and-place robot how to position the product.
“AI can help a vision system sort produce by color or cleanliness more thoroughly, or look at the animal carcass or larger cut of meat and analyze the best way to process it further down the line,” he adds. With AI applied to these systems, processors also can shift some of the human workforce out of the “dirty, dangerous or dull” jobs, allowing the machinery to handle quality control and sorting.
Building a foundation for AI
Although AI innovation appears to be charging fast and furiously, Izquierdo says that many in the food industry simply don’t have the right infrastructure in place to take advantage just yet.
“There isn’t so much a hesitation to implement the technology, but there are a lot of things processors need first,” he says. “There’s a lot of work still needed to get to a digital ecosystem collecting all the data, storing it, and sharing it in the cloud, and developing the right talent to analyze the data.”
Although many processors collect and analyze more data than they did in the past, it may not be enough to support AI. Furthermore, whether a processor works with a third-party data analyst or has hired an internal team, it’s still early in this data revolution for many in the industry.
“You need to have the whole digital ecosystem ready to take advantage of AI,” Izquierdo says. “Right now, rather than a full digital ecosystem, we have small islands in many companies looking for low-hanging fruit.”
In general, equipment manufacturers have jumped aboard the digital bandwagon, with most new equipment featuring the ability to provide a seemingly endless stream of data.
“The amount of data can be overwhelming these days, and if you have some older equipment installed on your lines, it’s getting easier and easier to retrofit to get the data,” he adds. The barrier to getting data has been lowered significantly in the last couple of years, and Izquierdo suggests companies continue to carefully lay the digital groundwork for AI.
“The right way to go is not to make everything digital, but instead to decide where you’re going to get the best payback,” he says. Keeping reality in focus and couching expectations in it is wise advice in the face of fast-paced innovation.
“In the past, many times we would all think about AI and view it as science fiction,” Izquierdo concludes. “But today, we think about these new systems that will help people write or edit — and food processors sort, analyze and process or package food products.”
The long-range possibilities for AI may still reside in humankind’s wildest dreams, but as innovation advances quickly, those dreams are fast becoming a reality that food processors need to embrace as quickly — and reasonably — as they can.
