While expectations of artificial intelligence are often at the extreme - a solution for everything versus the downfall of civilisation - the countless applications that have already been implemented and are bringing benefits tend to be overlooked. However, wherever data volumes exceed human processing capacity, AI is either already in use or should be. This is particularly true in the industrial environment. And therefore also when it comes to packaging.
In these five processes, artificial intelligence helps to make packaging more sustainable:
1. getting to the root: AI algorithms discover new materials
Remarkable news made the rounds in winter 2023. The AI tool GNoME from Google subsidiary DeepMind had discovered the structure of more than two million new crystals. These included around 380,000 stable materials that could actually be produced in the laboratory.
"We need new materials"
"To create a more sustainable future, we need new materials," says DeepMind. "GNoME has discovered stable crystals that have the potential to develop more environmentally friendly technologies." Although in this case AI research is more concerned with possible applications in superconductors or batteries, the direction seems clear: artificial intelligence is not only capable of designing new materials. It can also simulate their properties before they exist haptically.
32 million potential materials
The Pacific Northwest National Laboratory PNNL and Microsoft are also pursuing this approach: their collaboration recently led to the identification of 32 million potential materials for more sustainable batteries. Within a few months, the research team had selected the most suitable candidate and synthesised it for testing.
The prospect of more sustainable plastics
It is obvious that packaging materials have not been the primary target of this AI research to date. But this will not remain the case; the search for new materials with the help of AI algorithms has only just begun. Above all, the possibility of researching their properties virtually before production will lead to breakthroughs. Last year, researchers at Berkeley Lab in the US proved that the research and development of plastics is far from complete. They developed a plastic that can, in principle, be recycled an infinite number of times.
"The diversity of today's packaging plastics is almost limitless. Recycling these plastics back into high-quality plastics for packaging applications in the sense of a functioning circular economy is therefore a huge challenge and can hardly be solved sustainably without AI," says pacoon Managing Director Volker Muche.
2. at the beginning: artificial intelligence takes over the design
Industrial design and prototyping are often cited as core competences and key benefits of AI. Sceptics may search for images of AI-generated packaging in their search engine of choice. The results are astonishing. Apart from the visual appearance, however, the sustainability context is primarily about design that promises ecological progress. And here, too, artificial intelligence is capable of great things.
The worst mistakes happen at the beginning
Studies have amply demonstrated the importance of design in this respect. Around 80 per cent of the environmental impact of products can be traced back to decisions made in the design phase. This is just as true for smartphones as it is for packaging.
The technical side of packaging design is usually very conservative. If it is IT-supported at all, the basis for rigid packaging, for example, is usually the finite element method. FEM mathematically divides bodies into partial bodies such as cuboids. As their physical characteristics are known, the characteristics of the entire body can be deduced quite well.
Optimisation according to different parameters
AI solutions go far beyond this. They allow the design of packaging that is optimised in terms of wall thickness, structure, stability or general material savings in a way that neither human experience nor FEM can achieve. As is always the case with AI, this is based on corresponding databases, which in this case are filled with vast amounts of details and information on packaging.
AI-based packaging assessments
The approach also works retrospectively: AI solutions such as OptimAI, which pacoon uses, enable AI assessments of packaging in order to identify the optimisation potential of existing packaging. Experience shows that up to 30 per cent material savings or performance improvements can be achieved with the same amount of material. For example, by reducing the wall thickness or by remodelling the entire body.
This form of optimisation also takes into account existing filling processes in order to avoid or minimise changeovers. The system also offers the benefit of an automatic comparison of the weight, performance and eco-footprint of the packaging.
"Determining the optimisation potential of rigid containers, with or without iconic shape designs, using conventional methods is extremely time-consuming; AI develops a multiple of solution options in a fraction of the time," says Muche.
Packaging tests without packaging
But back to the visuals: the prototypes generated thanks to AI systems not only allow virtual testing of physical properties through to recyclability. They also offer the opportunity to obtain consumer insights via tests without having to produce haptic packaging. This saves material and energy, which has a positive impact on the packaging's footprint.
AI applications can also produce "designer pieces"
And anyone who thinks that artificial intelligence can optimise technology here, but is not able to create design in the aesthetic sense, should once again refer to their trusted search engine: The databases that the systems draw on are apparently also so well stocked in this respect that actual "designer pieces" are generated time and again.
3. to the machine: AI applications optimise production
What applies to industrial production in general also applies to packaging production and the packaging process in particular: they are among the best playgrounds for the use of artificial intelligence in the course of automation. Across the entire production line, AI helps to minimise waste - waste of energy first and foremost, but also of resources, time and manpower.
Condition monitoring and predictive maintenance
A classic example of this is predictive maintenance, one of the most established use cases for AI in production in a wide range of industries. The example of Mondi in Gronau, Westphalia, shows what predictive maintenance can achieve in the production of packaging. The company produces around 18 million tonnes of plastic and film products a year on around 60 extrusion, printing, gluing and winding machines. Each machine records 300 to 400 parameter values per minute, generating seven gigabytes of data per day.
Data that is used with the help of AI to anticipate impending machine failures in good time and plan maintenance accordingly. With convincing results: According to Mondi, the predictive maintenance of just eight of the machines is already generating savings of more than 50,000 euros per year.
Seamless quality control of packaging
Artificial intelligence also takes quality control to a higher level. Faulty packaging and subsequently damaged contents not only result in expensive compensation claims or even product recalls. They can also cause damage to the brand. AI-based quality control integrates visual inspection systems into the process that not only "see" defects, but also "understand" them thanks to intelligently processed data.
Learning from the images
The basis for this is training the AI with the help of images of intact or damaged packaging. As in all areas of AI application, a well-filled database is the prerequisite for satisfactory results.
Inspection systems of this kind make it possible to reject defective packaging almost completely. This applies to the production of the packaging itself as well as to the process of packaging goods.
4. at the workplace: AI technology saves on packaging
Anyone who frequently shops online may have noticed a change. Years ago, individual products were often delivered in cartons that were far too large and crammed with corresponding amounts of filling material. The fact that packaging sizes are now much closer to demand is also thanks to research and the use of AI solutions.
Intelligent customisation of product and packaging
The first step in parcel logistics, order picking, is a particular focus of automation and digitalisation. Just how seriously retail giant Amazon is taking the relevant research and products was recently demonstrated by its billion-euro investment in US start-up Anthropic. In addition to other AI hotspots in Amazon's warehouses, the focus is also on the packaging process.
In principle, the appropriate packaging size specified by the manufacturer is stored in the system for each product. However, if this is not the case, scanners are used to measure the length, width and height of the product and determine its weight. Information that either delivers the correct packaging to the packing station. Or, if the product does not fit into any standard packaging, it can trigger the cutting of a suitable box to size.
None of this is easy, as Volker Muche knows: "Packaging design and optimisation is a challenging cross-cutting issue, involving product protection, the supply chain, the filling process, material consumption, ecology and the consumer. The best compromise is optimal packaging."
Customers also have their say
Even customer feedback flows into the choice of packaging: artificial intelligence evaluates the recipient's reactions to the various packaging options. From 2015 to 2023, Amazon claims to have saved more than two million tonnes of packaging material with these and other measures.
The path to efficient sizing
It becomes one step more difficult when several items are ordered at the same time. The large online retailers also use AI systems here before dispatch, which use the data obtained from the scanners to calculate the optimum sizing of the products in the smallest possible boxes. This information is passed on to the pickers via monitors, VR glasses or other solutions.
5. to the end: AI development ensures more efficient recycling
AI already plays many roles in the various processes of waste and packaging recycling. In logistics, identification, separation and other steps. However, holistic approaches are also possible here.
When more than 50 partners from business, science and society come together to form innovation labs, the reason is likely to be an urgent one. The AI hub for plastic packaging is focussing on the last material step in the life cycle of packaging: recycling and the reuse of recyclates.
"As a packaging agency, we have a key function between this concentrated knowledge, material and technology expertise and the industry side with its applications. This interface expertise allows us to influence the coordination and cooperation of the value chain, which is where the true complexity of the issue lies," says Muche.
KIOptiPack: bringing more recyclates into the cycle
The KIOptiPack innovation laboratory is conducting research into the industrially safe use of recyclates. The aim is to develop practical AI-supported tools that - starting with product design - enable the production of plastic packaging with the highest possible proportion of recyclates. There is evidence that improvements in the utilisation of recyclates are necessary: Currently, only around eleven per cent of the volume is reprocessed into packaging. As the use of recyclates is made more difficult by factors such as fluctuating properties or the lack of chemically pure batches, KIOptiPack starts at the product design stage.
K3I-Cycling: closing the chain
The K3I-Cycling innovation lab starts with recycling. To this end, the team is using an artificial neural twin (ANT) to develop a new, open and standardisable AI interface that allows distributed process steps to communicate with each other. Process monitoring and process optimisation tasks are implemented decentrally.
The decisive factor here is that only data about the material flows is exchanged in the course of the research, not internal data. This is a key factor that should lead to cooperation and digital networking between all stakeholders in the value chain.
"How much AI will be in packaging in the future?"
Everyone is talking about AI. We work with it. We want to show you where AI can already be used today in the development and design of packaging and whether AI can also make packaging more sustainable.
Take part in our webinar! On 19 April from 11.00 to 11.45 am!
Click here to register