Innovation: AI in the supply chain: boosting tomorrow’s efficiency?
One example in e-commerce is the use of AI to accurately calculate package delivery routes and times. It can take multiple factors and parameters into account in real time (weather, peaks like Christmas, traffic conditions, historical patterns etc.), even if we have ordered something from the other side of the world. And in warehouses, it can be used to find the shortest distance between finding, picking and packing certain products.
Artificial intelligence: Conflicting goals in logistics
Hermes Newsroom recently interviewed Prof. Dr. Michael ten Hompel (TU Dortmund University) on this subject to get his perspective on AI in the supply chain. As a recognised expert in the use of new technologies in the broader logistics area, Prof. Dr. ten Hompel points out the complexity of the supply chain as one of the challenges. “It’s made up of multiple processes, companies, locations, people and technologies,” he says. “Logistics involves processing huge volumes of data while attempting to reconcile conflicting goals. That’s not such a problem if we want to optimise one process at a time, but it’s more of a challenge to optimise the entire chain.” He says the way it’s approached today is usually by using heuristics – rules of thumb (or even guesswork) based on historical experience. These may appear logical and may even work reasonably well, but are a long way from offering an optimal solution.
The Digital Continuum: interlinked and automated processes
Prof. Dr. ten Hompel’s work on logistical AI at Fraunhofer Institute and elsewhere involves developing simulations of logistical systems (as virtual reality) that comprise the specialised knowledge and intuition of highly experienced logistics specialists. “We’re working on the principle that what isn’t known can be learned,” notes ten Hompel. “By training AI to learn each step in an individual process, we can take things a step further and train it to understand a whole network of processes. AI agents can be taught to connect processes with each other, understand the relationships between them, detect anomalies and make decisions based on this information.”
The result is a closed loop of interdependent processes instead of a linear structure of standalone ones. “We call this the Digital Continuum – a seamless stream of automated processes that follow and interact with each other. In the supply chain, it enables processes to run faster and much more efficiently – across the manufacture, packaging, warehousing, transport, delivery etc. of a product – while significantly reducing logistics costs and bottlenecks,” adds ten Hompel. He says that the overarching goal of the Digital Continuum in logistics is to build a sustainable, resilient, adaptable and transparent supply chain.
Tomorrow’s world today: the LoadRunner
One example of how AI-based simulations can optimise logistics processes is the “LoadRunner”, developed by Fraunhofer Institute for Materials Flow and Warehousing in Dortmund, Germany. LoadRunners are small, autonomous vehicles that operate as a swarm. In other words, they communicate with each other like bees or birds do. The robots, which decide on their own trajectory, zoom around the warehouse at 10 metres per second, picking up and depositing packages while using their integrated cameras and communicating with each other to avoid collisions. For large loads, two or more LoadRunners can be physically attached using magnets. Two vehicles can also be connected wirelessly if they need to move together but be physically separate (such as to transport a long wooden beam). With the huge expansion of e-commerce that continues unabated, this type of technology saves both time and money. In tests at Fraunhofer Institute, 60 LoadRunners can collect and deposit around 10,000 packages an hour in the warehouse.
According to Prof. Dr. ten Hompel: “These vehicles can accelerate like a sports car and are pioneers for a whole new performance class. The LoadRunner is a central piece of the mosaic of tomorrow’s logistics.”
Against this backdrop, there’s a lot of exciting potential for AI in the logistics field. It will continue to evolve, delivering cost and efficiency benefits that ripple out throughout the supply chain.
About Prof. Dr. Michael ten Hompel
Michael ten Hompel holds the chair of Materials Flow and Warehousing at the TU Dortmund University, and is managing director of the Fraunhofer Institute for Material Flow and Logistics and Director of the Lamarr Institute for Machine Learning and Artificial Intelligence. In addition to his scientific activities, Michael ten Hompel has also been active as an entrepreneur. Among other things, he founded GamBit GmbH (now a subsidiary of Vanderlande Industries) in 1988 and managed the company, which primarily deals with the development and implementation of warehouse management systems, as managing partner until 2000. He is author and editor of numerous textbooks on logistics and IT and co-editor of Lecture Notes in Logistics (Springer). He is a board member of Bitkom. In 2012, he was elected to the Hall of Fame of Logistics and in 2017 he received an honorary doctorate from the University of Miskolc, Hungary.