With laptops on their knees or clutching red-and-black notebooks, a small group gathers every Friday at 10:00 a.m. on the dot, when Christian Menden holds an open session. “That’s an open exchange forum for colleagues who work on similar problems,” the 29-year-old explains. Today, as so often, the discussion revolves around the use of artificial intelligence – and specifically about process data forecasting for road haulage. What happens when logistical processes fail to work smoothly is familiar to anyone who has ever been stuck in a tailback on the freeway.
As part of the PRODAB logistics project, the ten employees from Menden’s department, who have backgrounds in business administration, statistics and mathematics, have lively exchanges of ideas about how to make planning processes more efficient in the supply chain management of customer projects. How do delays affect suppliers? What happens if employees fall ill? Or what if an elevator on the customer’s premises that is needed to carry goods breaks down? “Up to now, companies have scarcely been able to quantify the causes and effects of influences on and disruptions to logistical processes,” Menden says.
An economics graduate, Menden listens eagerly to his colleagues and repeatedly intervenes to dig deeper into various aspects. Later he will refer to this weekly get-together once again as a “very animated discussion.” Developing ideas in teams is “vitally important” to the Head of the Analytics Department at the Nuremberg location.
“People support each other, which helps them make faster progress, and every employee can actively initiate projects,” Menden says. “We identify future research topics and cooperation possibilities with a wide range of partners from industry.” Finding the right algorithm for each specific problem is the big challenge for him.
He explains how he developed a passion for applied research at an early stage. He was also attracted by the “dynamics of AI.” “What we want to do with AI is extract every last drop of efficiency, in particular when the data basis is poor.” The young scientist quickly established a career at Fraunhofer IIS – in a working environment that he loves. “As an institute with a strong mathematical focus, we combine forecasting and optimization methods with the goal of reducing the complexity of the mathematical modeling of industrial problems.”
Such as in the PRODAB project, where data is systematically collected for specific logistics processes using data analytics applications and modeled using Bayesian networks. And what are Bayesian networks, you may ask? As a specialist, Menden recognizes that the theorem, which originates in classical statistics, is an ideal method for AI compared to the frequently used neuronal networks, and leads to much greater efficiency. “If data isn’t available in sufficient quantities, it’s possible instead to integrate expert knowledge into Bayesian networks.” “If data isn’t available in sufficient quantities, it’s possible instead to integrate expert knowledge into Bayesian networks.” The software can then give specific recommendations for process improvement or the optimum allocation of resources,” Menden explains. His passion is, as he puts it, “to approach the mathematical optimum and carry out highly complex AI calculations, including in real time. For our customers, this means an improvement in resource usage and an increase in service level.”
Whether forecasting customer needs for production or inventory, or predicting spare parts requirements or freight volumes – which are often subject to fluctuations – or developing and launching precisely tailored health and motivation measures for warehousing personnel, “with such variety, there’s not much in the way of routine,” Menden says. And when wants to “zone out,” he seeks out one of the most attractive corners at Fraunhofer IIS’s Nuremberg location: the newly designed co-working space – with “the best coffee machine in the world.”
Written by Ilona Hörath