Artificial intelligence: challenges and best practices in mechanical engineering and intralogistics

On 23 May 2019, the 8^th Bavarian Innovation Congress on “Technological Innovations in Artificial Intelligence” was held at the Techbase in Regensburg. System Logistics GmbH gave the keynote presentation in front of a large audience including representatives from the political, business, educational and research communities. Several speakers from the Krones Group provided exciting insights into how artificial intelligence is already being used at Krones and what potential it offers for the future in mechanical engineering and intralogistics. One of the speakers was Dr. Patrick Glauner, Innovation Manager for Artificial Intelligence at Krones AG. I talked to him about the issue.

Some personal details to begin with: Dr. Patrick Glauner was awarded his doctorate at Luxembourg University for his thesis on detecting electricity theft by machine learning in developing and newly industrialising countries. He is also a lecturer on artificial intelligence at Karlsruhe University. He had previously worked for the European Organisation for Nuclear Research (CERN) and at the Université du Québec à Montréal. He studied computer science at Imperial College London and at Karlsruhe University, and also has a master’s degree in business administration.

Dr. Glauner, can you give me a short-and-sweet explanation of what artificial intelligence (AI) is?

There are multiple definitions for the term “artificial intelligence” (AI). I personally would define AI as follows: the aim of AI is to automate human behaviours in software and hardware. One vivid example here is the self-driving car, which navigates fully automatically, and thus imitates human behaviour.

How will AI transform our lives in the future?

AI will exert an enormous influence on our lives. Most of us nowadays are already coming into contact with it every day, without noticing it directly. Applications like Siri and Alexa, which process human speech, have long since become part of our daily lives. Spam filters, too, are a classic example of how AI derives statistical patterns from examples and takes decisions.

The workplace world, in particular, will be very significantly transformed by AI. For instance, some activities will become redundant, and will thus be replaced by AI. On the other hand, many new fields of activity will emerge – AI will become a fundamental skillset even outside the field of computer science, e.g. in mechanical engineering and intralogistics. There are numerous other examples of how AI will transform our lives – for instance, in future a diagnosis at your doctor’s will be arrived at more swiftly and more accurately using AI. The final decision, however, will still remain with a human being, in this case the doctor.

How will AI be implemented technically?

There are essentially two approaches for technical implementation of AI: in the case of expert systems, logic and rules are used in an attempt to describe causal connections and decision-taking options, so as to derive behaviours. In the case of machine learning, there is deliberately no attempt to describe details; rather, the computer recognises statistical patterns from sample data, and can apply these to other data in order to arrive at a decision. Expert systems were widely used in the past. For some years now, though, new AI applications have been predominantly implemented using machine learning, since complex problems simply cannot always be described completely in terms of logic and rules. My opinion, however, is that in future both these approaches will be progressively interlinked, so as to combine the best of both worlds. And this is precisely what we’re doing in the Krones Group.

Can you give us some practical examples of this inside the Krones Group?

Machine learning is already being used today for predictive maintenance. For instance, sample data can be used to detect when parts fail, enabling automatic predictions to be made as to when parts have to be replaced in a client’s production operation. We are meanwhile using machine learning for costing quotations, using features of customised machines sold in the past in order to estimate resource consumption – e.g. electricity at customised machines to be built in line with a client’s specific requirements. We use expert systems, by contrast, to describe in terms of rules how lines are built, thus helping to partially automate and speed up order processing.

What specific added values can AI offer for the Krones Group and its clients?

AI is relevant for all sections of the value added chain. Integrating all data unleashes enormous potentials for reducing costs, for example, avoiding errors and optimising resource consumption levels and waiting times.

In terms of AI, where do we stand in the Krones Group today, and where shall we be in five years’ time?

AI is already being implemented within the Krones Group and in clients’ projects. In the years ahead, more use will be made of AI at Krones both internally and externally in machines and lines, so as to optimise the clients’ value added chains.

Where do you see the greatest challenges for AI?

Europe has to invest significantly more in AI in order not to be left behind by China. Almost every business model will be massively transformed by AI. In China, both the politicians and the business community have taken this on board, and accordingly see AI as a great opportunity to become the technologically leading nation in the medium term. In this field, the AI innovations in China are already impressive today, but at the same time alarming as well. Here I can recommend the book “AI Superpowers: China, Silicon Valley, and the New World Order”.

What fascinates you personally about AI?

What interests me about AI is that we can use it basically to optimise things, to automate irksome activities, to speed up processes and reduce costs. In technological terms, what’s more, I find AI extremely exciting. Hitherto, we have been using best practices and heuristics in many procedures involving information processing, in order to solve problems. This can most definitely be optimised still further by AI. AI will certainly create an added value for people and make all our lives simpler, safer and more convenient.

Thank you for this informative interview!