Fritz Mayr is the founder and managing director of CIM GmbH, a leading provider of intralogistics software and warehouse management systems. In the following interview, we discuss the biggest themes of 2022: from artificial intelligence to the coronavirus, from the energy crisis to the move towards automation – plus the future of PROLAG World.
Mr Mayr, you’ve been working in the intralogistics software sector since 1985. So you’re very much an industry veteran. How have intralogistics systems changed over the years, would you say?
I’ll start with something that’s hardly changed at all: customer requirements. If I think back to the systems we delivered in the 1980s, the requirements were very similar to today’s systems. The main differences were that graphical user interface hadn’t been invented yet, significantly fewer algorithms were used and less data was transferred. Technologies were much slower and simpler. But even back then, manually operated stacker cranes were fitted with screens. The emergence of the internet at the beginning of the 1990s was definitely a major turning point. It took off properly in the early 2000s and allowed larger datasets to be transferred much faster. This enabled us to implement the first SaaS systems, a hands-off option where the customer logs in and instantly has a warehouse management system. That was in 2004.
Generally speaking, automation has been growing rapidly since about 2010 and there’s also been a sharp rise in warehouse automation. Particularly in terms of cost savings, fully automated warehouses are an attractive option for many companies nowadays. We’re currently seeing a great deal of investment in automation technologies. Another big difference is better user interface design. Software is easier to use nowadays, onboarding is faster and work processes have been streamlined – all thanks to simpler, more intuitive interfaces.
How will warehouse management systems be used differently in the future?
The master data is currently the basis for the logical decisions made by a WMS. It stores information on the requirements of an article along with its size, weight, data on which items tend to be picked at the same time and which picking methods are used. This data needs to be maintained efficiently so that the system can decide which storage location to use. which is the shortest path through the warehouse, and so on. All this information can be gleaned from the master data. Regularly maintaining and updating this data is extremely labour-intensive. But if it’s done properly, it can be a real boost to warehouse efficiency. We know from experience that master data maintenance is a weak point in many companies, however.
Artificial intelligence technologies will essentially make master data maintenance superfluous in the future. AI learns from its mistakes and maintains its own master data, so to speak. Through reinforcement learning, the system quickly gets to know the best location for each article. Artificial intelligence will eventually mean less administration and that systems will always intuitively do the right thing.
That all sounds slightly utopian...
Yes, maybe it is utopian. But that’s the goal. AI also uses algorithms, of course, but the programming methods are completely different and there’s a greater focus on simulation: an AI system first has to learn the basics using simulation technologies. The task is totally different and the results are not always instantly comprehensible. It gets better over time, of course. We need to remember that we’re just at the beginning. Although everyone’s talking about AI, the technology is actually still in its infancy.
Is artificial intelligence set to be as much of a game changer as the internet?
Yes, I’d certainly say you can compare the two. We’re not even aware of the change from the outside because the system works so smoothly. As things currently stand, we need to provide some input to make the system do what we want it to do: Perform warehouse generation, define ABC criteria, decide whether to work with or without batch management, multi-order picking – the data and workflows must be established before anything else happens. Which is the best picking method? One-step or two-step picking? This depends on the structure of the orders. These things are currently taken care of by a logistics consultant or a good administrator who knows the system well. They’re responsible for setting everything up. Intralogistics is so complex that it can’t function without an intelligent control system. Future systems will be able to do everything themselves. But it’s not something that can be achieved overnight. Classic programming implements logic which is based on analysis, whereas AI systems need to learn through experience. It’s a fascinating technology.
CIM is currently collaborating with AI researchers from the Technical University of Munich, among others. Are there any results yet?
A lot of the tests are looking very promising. The reinforcement learning appears to be working and the simulated intralogistics are becoming significantly more efficient. Artificial intelligence will certainly change the way modern warehouses conduct their business. Besides the ongoing project with the Technical University of Munich (TUM), we have two further research projects in the starting blocks. I’m very optimistic and proud that our team is at the cutting edge when it comes to practical application: We’ve already implemented our AI system in a customer test environment. Artificial intelligence is not just a technology of the future, in other words, it’s already part of everyday life here at CIM.