Future supply chains are multimodal and digitally organized

Time pressure and competitive pressure are increasing rapidly in logistics. Artificial intelligence (AI) and the intelligent use of digitalization technologies support the industry very efficiently with planning and decision-making tasks. This is demonstrated by numerous research projects in which scientists from Fraunhofer IML are involved.

AI support along the complete supply chain offers a great potential for increasing efficiency, as shown by numerous projects along the supply chain. For example, in the AutoModal project, a software program was developed for loading and unloading that monitors the surroundings of the handling crane and detects people as well as vehicles. This is the basis for automatic operation of the crane when handling containers in the trimodal terminal. The politically desired competitiveness of inland water transport is provided by the OKTOPUS project (Optimization of the Logistics and Scheduling Processes in the Maritime-based Transport Chain through Machine Learning in Steel Logistics), in which a forecast system was developed on the basis of artificial intelligence (AI) methods. This project is to predict transport times on the water of the Rhine corridor between Rotterdam and the Ruhr region accurately to a few minutes and determine capacity requirements in the case of low water.

When shipments arrive at an inland port, they are loaded onto trucks for further transport to a distribution center (DC). Detailed planning of the resources within the DC requires accurate prediction of the arrival time. However, not every company has the means and prerequisites to implement and use corresponding technology. Within the BMVI-supported project “Silicon Economy,” an open-source solution was developed for AI-based ETA (estimated time of arrival) prediction. This is based exclusively on open and freely accessible data sources and can also be used at less digitalized companies. To calculate the ETA, the AI uses, for example, weather forecasts that affect the driving speed, current and past traffic jam information, or, if necessary, legally required breaks, taking into account available parking spaces at truck stops. 

As soon as a shipment arrives at a distribution center, it has to be handled and prepared for local transport. This requires corresponding resources at the warehouse as well as in local transport. The corresponding resource planning can be greatly simplified and more accurately forecast with predictive analytics, a sub-area of AI. This task, which is currently often done manually, can be almost completely automated and helps to make work easier and save costs. The result of this “predictive analytics” approach are predictions that show future volumes much more accurately. 

When goods are on the last mile to the customer in a delivery vehicle, then the transparency and above all the punctuality of the delivery are what counts. Using AI methods, driving times and idle times can be learned from previous tours under a wide variety of conditions, which helps to increase the quality of tour planning. For this purpose, the geo-coordinates and the current speed are recorded in 30-second intervals on the driven tours. This data is used to develop a learning system that dynamically predicts the appropriate speed for individual road sections at specific times. For each journey segment, this results in corresponding time slices that predict the possible speed at the respective time. For optimum tour planning, the customerspecific idle times, in other words, the time from parking the vehicle to fulfilling the service and continuing the journey, are also of great interest. An AI-based expert system was therefore developed that predicts how long a driver will probably wait during a stop, depending on a multitude of different factors (e.g., number and weight of packages, season of the year and time of day, rural or urban region).

Due to their function as transshipment points, logistics hubs play an important part in supply chains. Holistic digital solutions can be created here using the Internet of Things (IoT) and AI-based approaches. The I2 PANEMA project, for example, uses nine demonstrators to show how processes at ports and thus in the supply chains can be made more efficient and secure. In particular, I2 PANEMA aims to use the possibilities of the IoT to improve port operations, make them more sustainable and thus pave the way for networks of intelligent ports. One key focus area in the project is the topic of IT security.

In the future, mutually compatible shipping documents for international road freight transport (CMR) will also be created electronically and thus stabilize supply chains. For this purpose, the eCMR project was launched within the Open Logistics Foundation Community. The eCMR uses the results of the eCMR project that was carried out at Fraunhofer IML in the framework of the “Silicon Economy.” The electronic consignment note is based on the creation, storage and transmission of digital consignment notes and machine-readable digital consignment notes, taking into account established templates and international standards. As a common data source, the eCMF also serves as an “enabler” for further digital processes, such as, e.g., automatic invoicing and payment. All developed components are made available to companies in the Open Logistics Foundation Repository.