Using AI to optimize packaging and processing

In an industry already heavily reliant on automation and robotics to optimize various supply chain and warehousing processes, how can packaging and processing firms continue to make efficiency gains?

At Packaging Innovations 2025, a panel comprising automation and sustainability experts tried to answer just that – with a key takeaway from the discussion being artificial intelligence’s (AI) growing role in the packaging sector, particularly in tandem with robotics.

Part of the show’s conference program, the panel discussion’s speakers comprised IMB sustainability and AI specialist Woody Falck; Mike Wilson, chief automation officer at the Manufacturing Technology Centre (MTC), Joe Muscat, environmental stewardship and innovation senior director at consumer health major Haleon, and Emmanuel Ewah, innovation lead at Innovate UK (moderator: Neil Farmer).

So how can the latest innovations make supply chains leaner?

Marrying AI and robotics

MTC’s Mike Wilson said the cost of automation has decreased over the years, making the technology more accessible. Capability has also improved.

“Robots are now designed to solve specific problems,” he opened. “Delta robots on the manufacturing line are designed in particular to give you very high speed. It’s nothing like it was used to be; it’s almost plug-and-play and you often see robots and Vision moving together.

“More recently, we’re seeing things like artificial intelligence being built into the robot systems. It’s already there, it’s actually helping increase the capability of these automation systems, at the same time making them easier to apply and to operate.”

IBM’s Woody Falck said: “AI is a very obvious approach to improving the supply chain. But at the minute, everyone thinks the answer is AI but don’t know what the problem is. And we need to step back and think, is this the most optimal way of doing things?”

“There’s a lot of usages for AI from the supply chain to R&D, [for example] reducing the number of errors on the assembly line. But then you have to be very specific in that AI isn’t always the answer – especially when it comes to sustainability.”

Cost constraints are also at play here, Wilson added. “Let’s say you’re looking to adopt AI as a business. How do you do that? One of the big things that means it matters is the cost constraints. AI, on the other hand, is extremely expensive to get together and develop,” he explained.

AI and waste reduction

Can the technology be used to cut down waste? Helion’s Joe Muscat explained the consumer health giant behind brands like Colgate is testing if an AI-powered model can help improve toothpaste recycling.

“One of our biggest projects [is] a recycling trial in New Jersey,” he said. “We installed a visual-based system, and trained it to detect all toothpaste tubes in realtime – not just Colgate, but all tubes.

“We’ve been running this for about a year now, collecting data, where in parallel, we’re trying to communicate with the wider value chain and community recycling programmes, and see whether that’s leading to an uptick in terms of recycling.”

The project is carried out in partnership with Mazza Recycling Services and Glacier.

The project demonstrates how AI can be used to pick out products that would traditionally end up in general waste streams rather than recycling.

“The challenge with lots of different materials coming along is that they’ve been maybe designed for recycling – but if a machine can’t detect products as a recycling item, then they’re just going to end up with residual waste and get out from the concentrated area,” Muscat said.

“There are some vision-based systems that go down the line and detect an old tube or a new bottle design, and based on the [AI model] training, you can also detect the types of plastic.

“Couple that with robotics, where you can pick these materials off, and that takes you to a conscious set of design and manufacturing data that can make a difference in terms of your products lifecycle.”

In the food industry, AI can be trained to enhance automation systems to reduce labor costs too – but there are caveats, according to IBM’s Falck.

“Dealing with large amount of products which are all different sizes in size and also very delicate is a big challenge,” he said. An example is a sandwich assembly line, where manual labor is still common and introducing AI to automate that task ‘is really difficult, because you’re just dealing with something that’s way too steep’, he explained. But technology that integrates robotics and AI to optimize production and reduce labor costs is starting to emerge, he added.

Getting started with AI integration

So where should packaging and processing businesses even begin? Whatever your requirements, start small, Muscat suggested.

“There is a learning curve, so find something simple to start with,” he said. “Start easy and you’ll be more successful – you’ll reduce risk, and once you’ve been through that learning curve, you can move on to the more challenging applications.”

Be mindful that any system specified today would be making yesterday’s pack, he added; so be flexible in anticipating the market in the next three to five years.

Either way, taking action now should be a priority.

“Businesses of all sizes, all industries are already doing this,” he said. “Everybody needs to be getting on the wave around how they can make their business more competitive.”